1 Timing, relations and cause of plutonic and volcanic activity of the 2 Siluro-Devonian post-collision magmatic episode in the Grampian 3 Terrane, 4 5 J.C. NEILSON1,3, B.P. KOKELAAR1 & Q.G. CROWLEY2,4 6 1Department of Earth and Ocean Sciences, University of Liverpool, Liverpool, L69 3BX, UK 7 2NERC Isotope Geosciences Laboratory, Keyworth, Nottingham, NG12 5GG, UK 8 3Present address: Dana Petroleum plc, 17 Carden Place, Aberdeen, AB10 1UR, UK 9 4Present address: School of Natural Sciences, Department of Geology, Trinity 10 College, Dublin 2, Ireland 11 12 (e-mail: [email protected]) 13 14 10340 text words; 119 references; 9 figures 15 Supplement: 445 words; 8 references; 1 table; 2 figures; 16 Grampian plutonic and volcanic activity 17 18 Abstract: Calc-alkaline magmatism in the Grampian Terrane started at ~430 Ma, 19 after subduction of the edge of continental Avalonia beneath Laurentia, and it then 20 persisted for ≥22 million years. ID-TIMS U-Pb zircon dating yields 425.0 ± 0.7 Ma 21 for the Lorn Lava Pile, 422.5 ± 0.5 Ma for Moor Pluton, 419.6 ± 5.4 Ma for 22 a Fault-intrusion at Glencoe volcano, 417.9 ± 0.9 Ma for Clach Leathad Pluton in 23 Glencoe, and in the Etive Pluton 414.9 ± 0.7 Ma for the Cruachan Intrusion and 408.0 24 ± 0.5 Ma for the Inner Starav Intrusion. The Etive Dyke Swarm was mostly emplaced 25 during 418-414 Ma, forming part of the plumbing of a large volcano (≥2000 km3) that 26 became intruded by the Etive Pluton and was subsequently removed by erosion. 27 During the magmatism large volumes (thousands of km3) of high Ba-Sr andesite and 28 dacite were erupted repeatedly, but were mostly removed by contemporaneous uplift 29 and erosion. This volcanic counterpart to the ‘Newer Granite’ plutons has not 30 previously been fully recognised. The intermediate magmas forming both plutons and 31 volcanoes originated mainly by partial melting of heterogeneous mafic-to-intermediate 32 lowermost crust that had high Ba-Sr derived from previous melting of LILE-enriched 33 mantle, possibly at ~1.8 Ga. This crustal recycling was induced by heat and volatiles 34 from underplated small-degree melts of LILE- and LREE-enriched lithospheric mantle 35 (appinite-lamprophyre magmas). The post-collision magmatism and uplift resulted 36 from breakoff of subducted oceanic lithosphere and consequent rise of asthenosphere. 37 Supplementary material: Geochronological methods, tabulated data and figures are 38 available at http://www.geolsoc.org.uk/SUP00000.

Neilson, Kokelaar, Crowley 1 39 40 Siluro-Devonian plutons populate a broad belt that extends ~300 km NE-SW across 41 the , on either side of the Fault. When intrusions of 42 similar age and setting in and Donegal are included (Fig. 1), the belt extends 43 to 750 km. The intrusions are the ‘Newer Granites’ of Read (1961) and are dominated 44 by the compositionally distinct and Northern Highland Suite (high Ba and Sr, 45 abundant appinites) and the Cairngorm Suite (relatively low Ba and Sr, few appinites) 46 (Stephens and Halliday 1984; Halliday et al. 1985; Tarney and Jones 1994; Highton 47 1999; Fowler et al. 2001). ‘Newer Granites’ in the Southern Uplands and in England, 48 comprising the Trans-Suture Suite of Brown et al. (2008), are younger than, and 49 unrelated to, those further north. Owing to contemporary uplift and erosion, remnants 50 of Siluro-Devonian volcanic rocks in the Highlands plutonic belt are relatively few 51 and mostly preserved south of the Great Glen Fault, in the ‘Grampian Terrane’ of 52 Gibbons and Gayer (1985). These include the centred or caldera volcanoes at Glencoe 53 (Kokelaar and Moore 2006), Ben Nevis (Bailey 1960) and Etive (Anderson 1937), a 54 thick and extensive sills-plus-lavas pile in Lorn (Kynaston and Hill 1908), and lavas, 55 tuffs and fossiliferous hydrothermal deposits at Rhynie (Trewin and Thirlwall 2002 56 and references therein). Small outliers of lavas and tuffs in the north, southeast and 57 south of the Grampian Terrane (Stephenson and Gould 1995, figure 26) indicate 58 originally extensive volcanic cover(s) on a broadly low-relief basement. 59 60 Northwest of the Great Glen Fault, some plutons older than ~425 Ma were involved in 61 late Scandian deformation: e.g. the 426 ± 2 Ma Strath Halladale Granite in the 62 Northern Highland Moines (Fig. 1; Kocks et al. 2006). Southeast of the Great Glen 63 Fault, in the Grampian Terrane, the intrusions and volcanic rocks are not pervasively 64 deformed, but are faulted and tilted; they respectively cut and unconformably overlie 65 Dalradian metamorphic basement. There seems consensus that sinistral strike-slip 66 movement on the Great Glen Fault, conceivably a minimum of 700 km, occurred from 67 ~428-410 Ma (e.g. Stewart et al. 1999, 2001; Dewey and Strachan 2005) and that this 68 resulted in the juxtaposition of Scandian-deformed intrusions northwest of the fault 69 with rocks showing no such deformation to the southeast. In the main Grampian belt 70 there was no crustal thickening or tectonic overprint due to the ≥435-425 Ma Scandian 71 orogeny, although there was uplift. 72 Fig. 1

Neilson, Kokelaar, Crowley 2 73 The petrogenesis of the ‘Newer Granites’ of the Highlands, their closely associated 74 appinites and lamprophyres, and the coeval hypabyssal and volcanic rocks, has 75 generally, but not universally, been considered to have involved hydrous mantle 76 melting above an active subduction zone that dipped beneath the Laurentian 77 continental margin (Dewey 1971; Brown 1979; van Breemen and Bluck 1981; Soper 78 1986; Thirlwall 1988; Fowler et al. 2001; Oliver 2001; Woodcock and Strachan 2002; 79 cf. Halliday and Stephens 1984). Some envisage an active continental destructive 80 plate margin analogous to the Andean Pacific margin (e.g. Oliver et al. 2008). Pitcher 81 (1982, 1987) and Watson (1984) believed that the ‘Newer Granites’ post-dated plate 82 collision – in the Southern Uplands they effectively stitch the Iapetus suture – and they 83 advocated a generalised origin involving decompression related to uplift and strike- 84 slip pull-apart. Pitcher (1982) emphasised that the Caledonian granitoids were unlike 85 those of the Andes, in both setting and composition. 86 87 The ‘Newer Granites’ of the Grampian Highlands are classified as I-type and high-K 88 calc-alkaline (Halliday and Stephens 1984), with enriched-mantle signatures and with 89 87Sr/86Sr initial ratios <0.708 (average 0.706). There has long been consensus that the 90 parental magmas were derived mainly from partial melting of metasomatised 91 lithospheric mantle and lower crust (Thirlwall 1982; Halliday 1984; Stephens and 92 Halliday 1984; Halliday et al. 1985; Tarney and Jones 1994; Fowler et al. 2001). This 93 contrasts with earlier intrusions, ≥434 Ma, which are of S-type, with relatively high 94 87Sr/86Sr initial ratios >0.712 (average 0.716), and are considered (e.g. Oliver 2001) to 95 result from uplift-decompression melting of crust (uplift ≥15-20 km) following the 96 Grampian orogeny. 97 98 Refinements in spatial and temporal reconstructions of the Caledonides (e.g. Stone et 99 al. 1987; Kneller 1991; Kneller et al. 1993; Dewey and Strachan 2003, 2005) cast 100 doubt that the Grampian magmatism was directly related to active subduction. In this 101 paper we distinguish a continuous ≥22 million-year magmatic episode that started just 102 when subduction would have been impeded through plate collision(s), which contrasts 103 with the discontinuous and protracted, syn-subduction assembly of the Andean 104 batholith (e.g. Pitcher 1978). We re-emphasise the point made by Pitcher (1982) that 105 the Caledonian plutons are unlike those of the Andean Cordillera; he referred to the 106 Grampian and Donegal plutons as ‘I- (Caledonian) type’ in recognition of the

Neilson, Kokelaar, Crowley 3 107 dissimilarity. In a multi-element (R1-R2) tectono-magmatic discriminant diagram 108 (Fig. 2a), the Grampian rocks do not plot with the Cordilleran I-type plutons in the 109 ‘pre-plate collision’ field, but in the ‘post-collision uplift’ and ‘late orogenic’ fields. 110 Also unlike the Andean magmatism (Pitcher et al. 1985), there was no precursor 111 extensional basin, no prolific extrusion of basalt, and no early plutonic phase of 112 gabbroic intrusions; a distinctive feature of the Grampian rocks is the paucity of 113 basalts (Figs 2b and c) and gabbros (Pitcher 1982). The lamprophyres and appinites, 114 however, mainly plot in the ‘pre-plate collision’ field. 115 Fig. 2. 116 Relationships between the Siluro-Devonian volcanoes, hypabyssal intrusions and 117 plutons are uniquely well displayed in the southwest Grampian Highlands (Fig. 3), 118 where cross-cutting relationships indicate protracted magmatism and progressive 119 assembly of a batholith. We take the Strath Ossian, Rannoch, Clach Leathad and 120 Etive plutons (Fig. 3) to constitute parts of a small batholith, named here 121 Batholith. Rannoch and Clach Leathad are both more extensive at depth than at 122 outcrop (Kokelaar and Moore 2006; this paper) and the plutons combine at depth to 123 form a continuous body with a southwesterly age progression. The plutons belong to 124 the Argyll and Northern Highland Suite (Stephens and Halliday 1984), although they 125 are not entirely of high Ba-Sr type (Fig. 2d). The extensive Etive Dyke Swarm, 126 dominantly of intermediate to silicic compositions (Fig. 2c), but including 127 lamprophyres, is related to a former centred volcano that was largely obliterated by 128 emplacement of the Etive Pluton and subsequent erosion (see below). The rocks 129 considered in this study are predominantly medium- or high-K calc-alkaline and 130 include monzodiorite-granodiorite-monzogranite-granite plutons, small-volume 131 appinite-lamprophyre intrusions, and volcanic suites of hypabyssal to extrusive 132 trachybasalt, basaltic trachyandesite, trachyandesite, trachyte and rhyolite, with 133 relatively sparse basalt, basaltic andesite and dacite (Figs 2b and c; after Le Maitre 134 1989); lavas and intrusions with shoshonitic affinity also occur. The prefix trachy- is 135 mostly dropped, for brevity. The Dalradian metamorphic basement in this region is 136 underlain by Proterozoic gneisses and mafic granulites, forming continental crust ~35 137 km thick. Basement rocks exposed to the west, on and Colonsay (Rhinns 138 Complex; Fig. 1), as well as magmatic rocks across the Grampian Terrane, indicate a 139 mafic crustal component with subduction-arc affinities derived from the mantle at 140 ~1.8 Ga (Halliday et al. 1985; Marcantonio et al. 1988; Dickin and Bowes 1991; Muir

Neilson, Kokelaar, Crowley 4 141 et al. 1992, 1994). Plutons widely across the region north of the Highland Boundary 142 Fault have long been known to carry inherited zircons with ages of ~1.6 Ga (Pidgeon 143 and Aftalion 1978). Nd, Sr and Pb isotope studies of the southwest Grampian rocks 144 show that the underlying lithospheric mantle is heterogeneous: probably layered with 145 tectonic interleaving and with variable metasomatism (Thirlwall 1982, 1986; 146 Macdonald and Fettes 2007; Neilson 2008). 147 Fig. 3 148 This paper aims to establish the timing and duration of the Grampian volcanic and 149 plutonic activity in the context of its tectonic setting, to constrain possible models for 150 the petrogenesis. Recent studies have considered a discrete pulse of plutonism, based 151 on poorly constrained age data (e.g. Atherton and Ghani 2002), or overlooked the 152 evidence for a probable block to subduction just when the I-type magmatism was 153 initiated (e.g. Oliver 2001; Oliver et al. 2008). We establish the importance of large 154 volcanoes and hence previously unrecognised large volumes of intermediate magmas 155 that bear on the interpretation of petrogenesis. Any petrogenetic model devised to 156 account for the ‘Newer Granites’ throughout the magmatic belt (excluding the 157 Southern Uplands and England), must also explain the generation of very large 158 volumes of andesite. 159 160 Temporal constraints on the cause of the magmatism 161 The main problem with linking the magmatism to subduction in the conventional 162 sense, i.e. with hydrous mantle melting due to dehydration of a down-going oceanic- 163 lithosphere slab, is in their relative timing. Here we review the subduction history, 164 before showing that magmatism in the southwest Grampian region was continuous for 165 at least 22 million years after continental collision. Northwards subduction of Iapetus 166 oceanic lithosphere began after the extended passive margin of Laurentia collided with 167 a substantial intra-Iapetus island arc during the early-mid Ordovician (Dewey and 168 Shackleton 1984; Dewey 2005). This collision was the Grampian event, when the 169 Dalradian strata were tectonically stacked and metamorphosed, between 478 Ma and 170 460 Ma. It involved a subduction-polarity reversal, from southwards beneath the arc 171 to northwards beneath it, towards Laurentia, and it resulted in the building of the 172 Southern Uplands tectono-sedimentary complex (Leggett et al. 1979; Kelling et al. 173 1987; Stone et al. 1987; Soper et al. 1999; Oliver 2001; Kinny et al. 2003; Dewey and 174 Strachan 2003, 2005; Dewey 2005).

Neilson, Kokelaar, Crowley 5 175 176 There is no record in the Grampian Terrane of subduction-related magmatism during 177 closure of the Iapetus Ocean. The terrane would have been several hundred 178 kilometres north of the subduction-trench while the contemporary arcs and associated 179 back-arc and fore-arc basins formed nearer the ocean, in the position of the Midland 180 Valley Terrane and further oceanwards, now buried beneath thrusted and back- 181 thrusted Southern Uplands strata (Stone et al. 1987; Heinz and Loeschke 1988; Bluck 182 1983, 2001). 183 184 Southwards subduction of Iapetus Ocean lithosphere beneath continental Avalonia 185 ceased in Caradoc times (~450 Ma). By late Llandovery times, ~430 Ma, the edge of 186 Avalonia had been subducted beneath the edge of Laurentia so that the Lake District 187 Terrane became the site of a foreland basin (Stone et al. 1987; Freeman et al. 1988; 188 Kneller 1991; Kneller et al. 1993). At this time, due to buoyancy constraints, 189 continued subduction of continental Avalonia would have been resisted; by ~428 ± 1.9 190 Ma sinistral strike-slip movement occurred along the Great Glen Fault (recorded by 191 the Clunes Tonalite; see Stewart et al. 2001). By mid-Wenlock times (~426 Ma) 192 subduction had slowed (Kemp 1987; Kneller 1991) and, in the late Wenlock, 193 sandstone turbidites of the M. lundgreni Biozone (Kemp 1987) were the last added to 194 the Southern Uplands complex. At the same time, detritus reworked from the 195 emergent complex extended far into the foreland basin, across the Iapetus Suture 196 (Furness 1965; Kemp 1987; Soper and Woodcock 1990; Kneller et al. 1993). By the 197 end of the Wenlock (~423 Ma), orthogonal shortening of the complex had switched to 198 sinistral transpression and a southeast-vergent fold-and-thrust belt developed in the 199 northern side of the foreland basin (Stone et al. 1987; Stone 1995). The thrust belt and 200 foreland basin migrated south through the late Silurian, with syndepositional 201 shortening during late Ludlow – Pridoli times, ~420-416 Ma. This migration has been 202 related to evolution of a northwest dipping mid-crustal thrust with southward 203 translation of an emergent (Acadian) mountain belt (Kneller and Bell 1993; Kneller et 204 al. 1993). We conclude that by ~430 Ma the supply of hydrated oceanic lithosphere 205 into the subduction zone had stopped and that further subduction of continental crust 206 was resisted. This was ~5 million years after ‘hard’ continental collision between 207 Baltica and Laurentia caused thrust and nappe deformation in Scandinavia, Greenland 208 and northwest Scotland (the Scandian orogenic event; Soper et al. 1992). We show

Neilson, Kokelaar, Crowley 6 209 below that I-type magmatism in the Grampian Terrane began at ~430 Ma, just when 210 subduction was resisted, and then continued for ≥22 million years. Magmatism 211 directly related to subduction might have persisted for a few million years after 430 212 Ma, but other processes must be invoked to explain the intense post-collision 213 magmatism that lasted at least 22 million years and produced a broad, linear belt of 214 plutons and volcanoes where none had formed previously. We agree with Atherton 215 and Ghani (2002) that an alternative to active subduction is required. 216 217 Uranium-Lead (U-Pb) zircon dating 218 U-Pb zircon dating was undertaken to facilitate understanding of the duration, timing 219 and rates of activity in the Siluro-Devonian magmatic episode. Previous studies have 220 mostly been limited and variable with regard to both the magmatic bodies studied and 221 the precision of the dating techniques. Thirlwall (1988) has provided the best 222 available regional synthesis of the timing of what he referred to as ‘Late Caledonian’ 223 magmatic activity, including Shetland, the Northern and Grampian Highlands, the 224 Midland Valley and the Southern Uplands; he considered all except that in the latter 225 terrane to be related to active subduction beneath Laurentia between 425 Ma and 408 226 Ma. Here we focus closely on a single element of that large area, with the benefit of 227 new plate-tectonic reconstructions and more precise dating techniques. The chosen 228 field area is where relative age relationships between various magmatic centres are 229 uniquely well constrained. Samples of one lava and five intrusions were selected: 230 Lorn Lava Pile, Rannoch Moor Pluton, Glencoe Fault-intrusion, Clach Leathad 231 Pluton, and the Cruachan and Inner Starav intrusions of the Etive Pluton (Fig. 3). The 232 Clach Leathad Pluton intrudes the Glencoe caldera volcano and has only recently been 233 recognised as separate from, and older than, the Cruachan Intrusion of the Etive 234 Pluton to the south (Kokelaar and Moore 2006; see below). The relationships indicate 235 that the Rannoch pluton is the oldest intrusion and Inner Starav the youngest; the age 236 relations of the Lorn Lava Pile to Glencoe and Rannoch were unknown. Appinites 237 and the Ballachulish Pluton have already been dated using U-Pb techniques, at ~429- 238 427 Ma (Rogers and Dunning 1991; Fraser et al. 2004). 239 240 Analytical Techniques 241 Zircon fractions were analysed by Isotope Dilution Thermal Ionization Mass 242 Spectrometry (ID-TIMS) at the NERC Isotope Geosciences Laboratory (NIGL),

Neilson, Kokelaar, Crowley 7 243 Nottingham, UK and at the Jack Scatterly Geochronology Laboratory, University of 244 Toronto. A full account of analytical protocols is given in the online supplementary 245 publication: http://www.geolsoc.org.uk/SUPXXXXX. 246 247 New U-Pb zircon ages 248 Thirty-four U-Pb zircon TIMS analyses were made on the six samples and are 249 presented in supplementary Table 1, with concordia diagrams in Figure 4 and final 250 results in Figure 5. Sample sites are located with 8-figure references to the Ordnance 251 Survey National Grid 100-km squares NM and NN. The geological context and 252 immediate implications of the new dates are discussed with each result; the broader 253 implications are considered afterwards. The new dates are entirely consistent with the 254 relative age relationships known from fieldwork. 255 Fig. 4 256 Fig. 5 257 Top of the Lorn Lava Pile 425.0 ± 0.7 Ma. Sample JNL86 is from the uppermost 258 dacite (Carn Gaibhre, NM 9740 2595; Fig. 5), near the top of the 600 m-thick lava 259 pile. It is purple and fine grained, with sparse phenocrysts of plagioclase and 260 hornblende, altered to sericite and chlorite respectively, and with minor apatite, 261 opaque oxides, zircon and titanite. The zircons picked for geochronology were 262 acicular to prismatic (200x70x70 μm to 120x50x50 μm), colourless and translucent, 263 commonly with magmatic c-axis inclusions (Supplementary Fig. 1a). A range of 264 zircon morphologies and sizes exists in this sample, including crystals with visible 265 cores (Supplementary Fig. 1b). Due to the small size of acicular zircons that were 266 picked, to avoid inherited cores, multi-grain fractions were generally used. Z-2 and Z- 267 6 yield a concordia age of 425.0 ± 0.7 (concordance and equivalence MSWD 1.9, 268 probability 0.13) and mean 206Pb/238U age of 425.0 ± 0.6 Ma (MSWD 2.5, probability 269 0.11), while Z-1 and Z-5 yield an older concordia age of 430.6 ± 1.3 Ma. Regressing 270 all four gives a lower intercept age of 425.5 ± 4.7 Ma. The 425.0 ± 0.7 Ma age is 271 considered to represent the true eruption age, with the apparently older zircons 272 involving an inherited component. It is noteworthy that although an apparent age 273 range of ~5 million years occurs in these Lorn dacite zircons, the analyses are 274 concordant suggesting that individual fractions either represent single-age populations, 275 or at least populations so close in age that the analyses are not discordant. The former 276 is confirmed by Cathode Luminescence (CL) imaging (Supplementary Fig. 1a); the

Neilson, Kokelaar, Crowley 8 277 analysed acicular zircons with c-axis melt inclusions do not have any inherited cores, 278 nor do they show any resorption textures. Accumulation of the Lorn lava pile would 279 have taken ≤1 million years, consequently the onset of the eruptions is placed at no 280 more than ~426 Ma; there are no known significant unconformities within the pile. 281 Previous (Rb-Sr) ages for Lorn include a lava at 423.5 ± 6.1 Ma and, from the Ferry 282 Plug intrusion southwest of Oban (Fig. 5), a range from 413.6 ± 6 to 426.9 ± 5 Ma 283 (Thirlwall 1988). 284 285 The Lorn Lava Pile is underlain by conglomerates that rest unconformably on the 286 Dalradian. They contain abundant clasts of quartzite, boulders of andesite (≤2 m) and 287 biotite granite (≤1 m), and cobbles or pebbles of quartz-porphyry, granodiorite and 288 granite. The clasts suggest that either there was earlier andesite volcanism somewhere 289 in the palaeo-Great Glen drainage system to the northeast (palaeocurrents were 290 towards southwest; Morton 1979; Neilson 2008), or the andesitic boulders were 291 derived from the developing Lorn pile, having been reworked into a canyon that cut 292 down to the Dalradian basement. The large size of the boulders suggests either that 293 they were locally derived or that the fluvial drainage was very steep. The 294 palaeocanyon or valley floor where the boulder conglomerates are preserved shows no 295 evidence for steepness; associated sandstones and siltstones record tranquil overbank 296 and shallow lacustrine sedimentation. The provenance of the silicic igneous clasts is 297 uncertain, but plutons were emplaced in the general vicinity, up-palaeovalley, and 298 these may have been unroofed and eroded before the influx of sediments: e.g. Clunes 299 Tonalite 427.8 ± 1.9 Ma (Stewart et al. 2001) and Ballachulish Igneous Complex 427 300 ± 1 Ma (Fraser et al. 2004). At the time of emplacement of the Ballachulish Pluton 301 there was 9-10 km of cover above the level of the present outcrop (Pattison and Harte 302 1997), but a shallow upper part may have been exhumed within as little as ~2 million 303 years (e.g. Harayama 1992), as later with the Rannoch Moor Pluton (see below). The 304 appinites at Garabal Hill (Nockolds 1941) and at Rubha Mhor near Ballachulish 305 (Bowes and Wright 1967) had also been emplaced (~429-427 Ma; Rogers and 306 Dunning 1991), but no such rock has been recognised in the Lorn conglomerates. The 307 boulders show that pluton unroofing, and possibly also andesitic eruptions, predated 308 426-425 Ma. The interval required for (rapid) unroofing places the age of plutonic 309 intrusion at no younger than 428-427 Ma, consistent with Ballachulish as a potential 310 source.

Neilson, Kokelaar, Crowley 9 311 312 Two silicic ignimbrites within the Lorn Lava Pile, east of (NM 91 37), 313 were tentatively linked by Roberts (1963) to eruptions from the Glencoe caldera 314 volcano. However, explosive silicic eruptions there started some 5 million years after 315 accumulation of the Lorn pile (~420 Ma; see below) and there is no obvious source for 316 the ignimbrites. As strike-slip on the Great Glen Fault continued long after 317 emplacement of the Lorn pile, it is probable that some substantial counterpart of the 318 pile northwest of the fault has been removed. The minor pipe-like intrusions on 319 southern Kerrera and to the south of Oban (Fig. 5) are not major feeders of the Lorn 320 pile and here the lowermost sheets are sills that were emplaced into unlithified wet 321 sediments (Neilson 2008). The Kilmelford intrusive suite, south of the Lorn outcrop 322 (Fig. 3), is dominated by biotite-hornblende diorites and granodiorite, and has 323 shoshonitic affinity (Zhou 1987) similar to Lorn (see below). Conceivably Kilmelford 324 could represent a plutonic counterpart to the Lorn extrusive pile, as suggested by 325 Tarney and Jones (1994). 326 327 The sedimentary strata beneath the Lorn lavas and sills around Oban and on Kerrera 328 have been renowned for their fossils of fish, eurypterids, millipedes, ostracods, plant 329 remains and sporomorphs, which have been assigned to an uppermost Silurian or 330 lowermost Devonian age (Lee and Bailey 1925; Morton 1979; Marshall 1991; 331 Wellman 1994; Wellman and Richardson 1996). The new radiometric age for the 332 Lorn lavas suggests that their sedimentary substrate is of an older Silurian age than 333 originally assigned, implying earlier occurrences for the flora and fauna; 426-425 Ma 334 is in the Homerian Stage of the Wenlock (Gradstein et al. 2004). 335 336 Rannoch Moor Pluton 422.5 ± 0.5 Ma. Sample JNGC88 is a foliated granodiorite 337 from the main body of the pluton (G2 of Jacques and Reavy 1994), 1.5 km from the 338 outer contact (NN 2610 5570; Fig. 5). It contains hornblende and biotite, with minor 339 apatite, titanite, opaque oxides and zircon. The analysed zircons were single-grain 340 fractions, acicular or squat (250x90x90 μm to 300x20x80 μm), colourless and 341 translucent. Four analyses [Z-5, Z-6, Z-8 and Z-9] define a concordia age of 422.5 ± 342 0.5 Ma (concordance and equivalence MSWD 0.88, probability 0.52) and a mean 343 206Pb/238U age of 422.7 ± 0.4 Ma (MSWD 0.67, probability of fit 0.57); the consistent 344 data are taken as the true crystallization age.

Neilson, Kokelaar, Crowley 10 345 346 The Rannoch crystallisation age of ~422.5 Ma post-dates the Lorn eruptions by some 347 2.5 million years and pre-dates the onset of centred volcanism at Glencoe. Boulders 348 of foliated Rannoch Moor granodiorite in conglomerates beneath the basal ignimbrite 349 at Glencoe show that the Rannoch pluton was unroofed before the onset of Glencoe 350 caldera volcanism (Kokelaar and Moore 2006). Older conglomerates at Glencoe, 351 beneath the Basal Andesite Sill-complex, contain granite, granodiorite and ‘basic 352 plutonic rock resembling kentallenite’ (Bailey and Maufe 1916), and, on the 353 reasonable assumption that fluvial drainage was from the east or southeast along the 354 Glencoe Lineament (Kokelaar and Moore 2006), the Rannoch Moor Pluton is the best 355 contender as their source. The boulders would have been derived from levels of the 356 pluton shallower than those presently exposed. Assuming a minimum depth of some 357 1.5-2 km to the top of the pluton and 2 million years as the minimum time for 358 unroofing (Harayama 1992), the conglomerates could only have been deposited at or 359 after 420.5 Ma. This is the probable maximum age of the andesite sills and certainly 360 the maximum age of the (somewhat later) onset of the caldera volcanism. Flora and 361 sporomorphs (Kidston and Lang 1924; Wellman 1994) from siltstones overlying the 362 conglomerates were considered to be early Devonian (late-early or early-late 363 Lochkovian) in age, which, from the timescales of Gradstein et al. (2004) and 364 Kaufmann (2006), would be ~415 Ma. However, 415 Ma is the age obtained for the 365 Cruachan Intrusion of the Etive Pluton (Fig. 5), which post-dates both the Clach 366 Leathad Pluton (~418 Ma) and the Glencoe volcano (~419 Ma). Considering the 367 radiometric ages from Rannoch and Glencoe, with the time required to unroof 368 Rannoch and derive the boulders, a ~420 Ma age for the fossiliferous siltstones and 369 onset of centred volcanism at Glencoe is indicated. 420 Ma is in the Ludfordian Stage 370 (Ludlow) of the Silurian (Gradstein et al. 2004). 371 372 Glencoe Fault-intrusion 419.6 ± 5.4 Ma. Sample JNGC76 is from the Gleann 373 Charnan Fault-intrusion in the southwest of the Glencoe Caldera-volcano Complex 374 (NN 1354 5103; Fig. 5). It is a hornblende-biotite quartz diorite, with minor apatite, 375 titanite, opaque oxides and zircon. The zircons are typically acicular (120x90x50 μm 376 to 100x80x60 μm), colourless and translucent. With the exception of Z-8, U-Pb data 377 for this sample plot on a Pb-loss trajectory from ~419 Ma; regression through the data 378 yields an upper intercept age of 419.6 ± 5.4 Ma (MSWD 1.01). Glencoe Fault-

Neilson, Kokelaar, Crowley 11 379 intrusions widely show evidence of hydrothermal alteration (e.g. Garnham 1988; 380 Kokelaar 2007), which may have caused extensive Pb loss. 381 382 This intrusion was selected for analysis because it is one of the few with an inferred 383 extrusive counterpart within the Glencoe Volcanic Formation. It probably represents a 384 feeder of the Bidean nam Bian Andesite Member (Kokelaar and Moore 2006; Neilson 385 2008). This member lies close to the top of the preserved volcanic pile and records a 386 single intra-caldera eruption of compositionally distinct batches of andesitic and 387 dacitic magmas totalling ≥12 km3. Thirlwall (1988) obtained a 40Ar/39Ar plateau age 388 of 421 ± 4 Ma from hornblende separated from this member. On the basis of the age 389 of the onset of caldera volcanism, at ~420 Ma (see above), the upper intercept age of 390 419.6 ± 5.4 Ma is regarded as the emplacement age for the Gleann Charnan Fault- 391 intrusion. Z-8 gives a 206Pb/238U age of 421.4 ± 0.7 Ma, which is interpreted as 392 recording inheritance, because it is highly unlikely that unroofing of the Rannoch 393 Moor Pluton and subsequent development of much of the Glencoe caldera-volcano 394 succession could have occurred in ~1 million years. Furthermore, the intercept age of 395 ~419 Ma is more readily reconciled with the ~418 Ma of the Clach Leathad Pluton 396 (Fig. 5). This pluton intrudes the youngest preserved part of the Glencoe Volcanic 397 Formation and is not chilled in contacts with a (different) fault-intrusion, in the 398 southeast. The evolution of the Basal Andesite Sill-complex and centred caldera 399 volcano at Glencoe, up to the time of its intrusion by the Clach Leathad Pluton, 400 occurred within ~2.5 million years. The unconformity between the Basal Andesite 401 Sill-complex and the overlying ignimbrite (Lower Etive Rhyolite) represents 402 approximately 0.5 million years (Kokelaar and Moore 2006), so that caldera 403 volcanism up to emplacement of the Clach Leathad Pluton lasted a maximum of 2 404 million years. During this time there were five large effusions of andesite-dacite and 405 at least seven caldera-forming eruptions of rhyodacite-rhyolite (Moore and Kokelaar 406 1998). 407 408 Clach Leathad Pluton 417.9 ± 0.9 Ma. Sample JNE5 was collected from the summit 409 of Clach Leathad (1099 m; NN 2470 4910; Fig. 5) in the main body of the pluton. It is 410 a pink and grey, hornblende-biotite monzogranite, containing minor apatite, titanite, 411 opaque oxides and zircon. The zircons display various morphologies, but those 412 analysed were either acicular or squat (220x80x60 μm to 280x120x100 μm),

Neilson, Kokelaar, Crowley 12 413 colourless and translucent. Five analyses [Z-27, Z-28, Z-29, Z-30 and Z-31] yield a 414 concordia age of 417.9 ± 0.9 Ma (concordance and equivalence MSWD 2.7, 415 probability 0.01) and a mean 206Pb/238U age of 417.8 ± 0.9 Ma (MSWD 3.5, 416 probability of fit 0.01), interpreted to represent the time of crystallization. 417 418 This date of ~418 Ma is the accepted age of the Silurian-Devonian boundary 419 (Kaufmann 2006) and is when magmatic activity centred at Glencoe ceased. It marks 420 the upper age limit of the subsequent protracted emplacement of the Etive Dyke 421 Swarm. Mineral alignment fabrics in the Clach Leathad monzogranite adjacent to the 422 earliest Etive dykes show that the latter were emplaced during advanced solidification 423 of the pluton, before its full crystallisation (Jacques 1995, p. 356). 424 425 The Clach Leathad Pluton was described (e.g. Bailey 1960) as a northern lobe of the 426 Cruachan Intrusion, and hence as part of the Etive Pluton to the south. However, field 427 relationships between the plutonic intrusions and with dykes of the Etive swarm led 428 Kokelaar and Moore (2006) to propose that the Clach Leathad Pluton is a separate 429 intrusion, centred at Glencoe and substantially older than the Cruachan Intrusion. This 430 reinterpretation is justified by the Clach Leathad Pluton proving to be ~3 million years 431 older than the Cruachan Intrusion (Fig. 5). The intervening time was when the centred 432 Etive Volcano developed (see below). 433 434 Cruachan Intrusion (Etive Pluton) 414.9 ± 0.7 Ma. Sample JNE8 is from the main 435 body of the intrusion, 800 m from its outer contact, in the region of Glen Ure (NN 436 0710 4740; Fig. 5). It is a pale grey hornblende-biotite granodiorite with minor 437 apatite, titanite, opaque oxides and zircon. The analysed zircons were either acicular 438 or squat (250x60x50 μm to 310x110x80 μm), colourless and translucent. Four 439 analyses [Z-33, Z-36, Z-37 and Z-38] yield a concordia age of 414.9 ± 0.7 Ma 440 (concordance and equivalence MSWD 2.6, probability 0.012) and a mean 206Pb/238U 441 age of 414.7 ± 0.7 Ma (MSWD 3.9, probability of fit 0.01), with the data taken to be 442 the best estimate of the true crystallization age. Two further [Z-34 and Z-35] analyses 443 are discordant due to Pb loss. 444 445 Inner Starav Intrusion (Etive Pluton) 408.0 ± 0.5 Ma. Sample JNGC87 is from near 446 the centre of the intrusion, in the vicinity of Ben Starav, >1 km from the contact with

Neilson, Kokelaar, Crowley 13 447 the Porphyritic Outer Starav Intrusion (NN 1375 4310; Fig. 5). It is a biotite 448 monzogranite, with minor apatite, titanite, opaque oxides and zircon. The analysed 449 zircons were either acicular or squat (240x90x90 μm to 150x90x80 μm), colourless 450 and translucent. Three analyses [Z-10, Z-41 and Z-42] yield a concordia age of 408.0 451 ± 0.5 Ma (concordance and equivalence MSWD 0.12, probability 0.99) and a mean 452 206Pb/238U age of 408.0 ± 0.3 Ma (MSWD 0.11, probability of fit 0.89), taken as the 453 crystallization age. One fraction [Z-4] is discordant due to minor-Pb loss while 454 another [Z-5] indicates some inheritance. 455 456 The duration of assembly of the Etive Pluton is defined by the age of the Inner Starav 457 Intrusion at 408.0 ± 0.5 Ma and some age between those of the Clach Leathad Pluton, 458 417.9 ± 0.9 Ma, and the Cruachan Intrusion, 414.9 ± 0.7 Ma. The earliest component 459 of the Etive Pluton to cut the Clach Leathad Pluton is the Meall Odhar Intrusion (Figs 460 5 and 6; Kokelaar and Moore 2006) against which the Cruachan Intrusion shows no 461 chilled margin. Consequently the age for the Meall Odhar Intrusion is no older than 462 ~416 Ma. Thus the Etive Pluton was assembled in ~8 million years. During this time 463 the Lintrathen ignimbrite was emplaced in the southeast of the Grampian Terrane, at 464 ~415 Ma, and volcanic activity with hydrothermal mineralisation formed the 465 renowned plant-bearing Rhynie chert, at ~412 Ma (Trewin and Thirlwall 2002; S. 466 Parry pers. comm. 2008). 467 Fig. 6 468 The Inner Starav Intrusion was the last major body of magma of the southwest 469 Grampian suite emplaced in this region. 408 Ma is in the late Pragian (Gradstein et al. 470 2004) or early Emsian stage (Kaufmann 2006) of the Lower Devonian, after which 471 there was folding in the Midland Valley and development of a widespread 472 unconformity in Scotland. The magmatism of the Trans-Suture Suite of ‘Newer 473 Granites’ in the Southern Uplands was initiated at ~400 Ma (Brown et al. 2008), at 474 least 5 million years after the main activity in the southwest Highlands had ended. 475 476 New ages recently published by Oliver et al. (2008) and concerning rocks that we 477 have analysed are in conflict with the geological field relationships. Their data, 478 derived from sensitive high-resolution ion microprobe (SHRIMP) U-Pb analyses of 479 zircons, place Rannoch Moor Pluton at 409 ± 8 Ma (in contrast to our 422.5 ± 0.5), 480 which would make it one of the youngest elements of the southwest Grampian suite,

Neilson, Kokelaar, Crowley 14 481 despite the fact that Rannoch is cut by Etive dykes that predate the Outer Starav 482 Intrusion of the Etive Pluton, which they have dated at 415 ± 6 Ma. Also, ages 483 published by other authors and utilised by Oliver et al. (2008), for example the 406 ± 6 484 and 398 ± 2 Ma respectively for rhyolite and andesite at Glencoe (Fraser et al. 2004), 485 are similarly impossible to reconcile with the geological field evidence and with our 486 new results (Glencoe ~420.5-418 Ma; see above). 487 488 Nature and timing of appinite-lamprophyre magmatism 489 Appinites and high-K calc-alkaline lamprophyres, respectively coarse-grained and 490 porphyritic varieties of incompatible-element-rich rocks with abundant hydrous mafic 491 phases and commonly some K-feldspar, ultimately derive from small-degree partial 492 melts of enriched (K-amphibole and/or phlogopite-bearing) sub-continental 493 lithospheric mantle (e.g. Rock 1991; Fowler and Henney 1996). That kentallenite 494 (phlogopite and K-feldspar-bearing picrite), appinite and lamprophyre intrusions are 495 genetically related was first recognised in the southwest Grampian region, by Hill and 496 Kynaston (1900). The high contents of incompatible trace elements (especially Ba 497 and Sr) in these rocks, and in some mafic enclaves and dioritic components in plutons 498 belonging to the Argyll and Northern Highland Suite, have long been taken to indicate 499 a petrogenetic link (e.g. Stephens and Halliday 1984; Macdonald et al. 1986; Rock and 500 Hunter 1987; Atherton and Ghani 2002). Several authors have suggested that high 501 Ba-Sr ‘Newer Granites’ may derive from the mantle-derived melts primarily by 502 fractional crystallisation with crustal assimilation (e.g. Rock and Hunter 1987; Fowler 503 et al. 2001). That syenite and granite can derive from fractional crystallisation of 504 lamprophyric magma has been demonstrated in a dyke (Macdonald et al. 1986) and 505 interpreted in hybrid appinite-syenite-granite complexes (Fowler 1988; Fowler and 506 Henney 1996). 507 508 The southwest Grampian ‘Appinite Suite’ was originally considered to have been 509 emplaced during a relatively short time, generally before intrusion of the ‘late 510 Caledonian’ granites (Read 1961; Wright and Bowes 1979; Rogers and Dunning 511 1991). More widely in the magmatic belt, the appinites are thought to have been 512 intruded before or during emplacement of the granites with which they are 513 (physically) associated (e.g. Pitcher and Berger 1972; Fowler et al. 2001; Atherton and 514 Ghani 2002; Macdonald and Fettes 2007). In the southwest Grampian region we find

Neilson, Kokelaar, Crowley 15 515 that small-degree mantle melts that would form appinite or lamprophyre were 516 available through much of the ≥22 million-year magmatic episode, although 517 seemingly their arrival in the upper crust was most common during early stages. 518 519 The oldest U-Pb zircon age for magmatic crystallisation in the ‘post-collision’ rocks of 520 the southwest Grampian region is 429 ± 2 Ma (Fig. 5; Rogers and Dunning 1991), 521 from Garabal Hill appinitic diorite in the Glen Fyne igneous complex (Nockolds 522 1941). This age was derived from three discordant zircon analyses and the ± 2 Ma 523 error seems low. Re-evaluation of the data gives a 428 ± 9.8 Ma upper intercept age 524 derived from a discordia through the three zircon fractions. Air abrasion (Krogh 525 1982) of these fractions probably failed to remove all of the effects of Pb-loss. It is 526 likely that volcanism started before plutonic crystallisation, which would push the 527 initiation of magmatism perhaps to ~430 Ma. 528 529 The Rubha Mor appinitic intrusion (427 ± 3 Ma; Rogers and Dunning 1991) was 530 considered, as was the type kentallenite nearby, to have been emplaced before the 531 adjacent Ballachulish Pluton (Fig. 5). However, the subsequent determination of 427 532 ± 1 Ma for that pluton (Fraser et al. 2004), which has gradational contacts with other 533 appinites (Weiss and Troll 1989), suggests that the Rubha Mor and Ballachulish 534 intrusions were practically contemporaneous. The original 427 ± 3 Ma age was 535 derived from a single titanite fraction and re-evaluation of the data gives the Rubha 536 Mor titanite a 206Pb/238U age of 426.5 ± 4.2 (2σ, includes decay constant errors). 537

538 Shoshonitic lavas (SiO2≤57.0 wt.%; K2O≥Na2O-2.0) that are petrographically and 539 compositionally indistinguishable from lamprophyre, and minor lamprophyre 540 intrusions, occur in the Lorn Lava Pile (see below), dated here at 426-425 Ma. 541 542 Magma-mixing and mingling relationships between the Clashgour appinitic intrusion 543 and its host monzogranite indicate the co-existence of the contrasting magmas here 544 (Supplementary Fig. 2). The Clashgour outcrop (Figs 3 and 5) is most probably a part 545 of the Rannoch Moor Pluton, which is dated here at 422.5 ± 0.5 Ma, but it could be 546 part of the younger Clach Leathad or Etive plutons. Thus magmas that formed 547 appinite were available for at least 6 million years in the early part of the episode. 548

Neilson, Kokelaar, Crowley 16 549 Lamprophyre dykes are present widely within the Etive Dyke Swarm (e.g. Kynaston 550 and Hill 1908), which formed from 418 Ma onwards for more than 3 million years 551 (see below). One lamprophyre dyke cuts the contact between the 415 Ma Cruachan 552 Intrusion and the younger Outer Starav Intrusion, and is chilled against both (Bailey 553 and Maufe 1916, p. 124), and another, thin and amygdaloidal, lamprophyre dyke 554 occurs close to the middle of the 408 Ma Inner Starav Intrusion (Fig. 5; 1.4 km south- 555 southwest of the summit of Ben Starav). The latter contains phenocrysts of olivine 556 (pseudomorphs), augite, hornblende and phlogopite, with K-feldspar, plagioclase, 557 apatite, magnetite and calcite (see also Kynaston and Hill 1908, p. 117) and it might 558 record the last phase of Siluro-Devonian magmatism in the region. Thus, mafic small- 559 degree partial melts of lithospheric mantle that would form appinite or lamprophyre 560 were present through at least 15 million years, ~430-<415 Ma, and possibly 561 throughout the ≥22 million-year post-collision magmatic episode. 562 563 Etive Dyke Swarm and associated Etive Volcano 564 The Etive Dyke Swarm extends over 100 by 20 km, with the Etive Pluton near its 565 centre (Fig. 3). Although the swarm is well known, it has not been well understood; 566 there has been a tendency (e.g. Morris et al. 2005) to stress a relationship with the 567 developing Etive Pluton while overlooking evidence that the swarm was well 568 populated with dykes before the plutonic intrusion at this level. Here it is proposed 569 that the swarm originated from the plumbing of a developing centred volcano that 570 eventually became intruded by the pluton and was subsequently mostly removed by 571 uplift and erosion. 572 573 Analyses of Etive dykes are plotted in Figure 2c and d. Both compositionally and 574 petrographically the (fine-grained) dykes are indistinguishable from southwest 575 Grampian volcanics, being closely matched by the diverse rocks of the earlier- 576 emplaced Lorn Lava Pile. Numerous authors, e.g. Anderson (1937), following Bailey 577 and Maufe (1916) and Kynaston and Hill (1908), have distinguished ‘felsite and 578 porphyry’: dacites and rhyolites with or without quartz and feldspar phenocrysts, 579 ‘prophyrites’: andesites and dacites with phenocrysts of hornblende or biotite and 580 plagioclase, ‘microdiorites’ and ‘lamprophyres’. No systematic compositional 581 sequence through time has ever been properly established and, by analogy with Lorn 582 (see below), no systematic change need be expected.

Neilson, Kokelaar, Crowley 17 583 584 As Etive dykes intrude the Cruachan Intrusion (Fig. 6), it is evident that the main 585 development of the Etive Dyke Swarm took more than 3 million years, between the 586 cooling of the Clach Leathad Pluton at ~418 Ma and after crystallization of the 587 Cruachan Intrusion at ~415 Ma. Few dykes intrude the more central components of 588 the Etive Pluton, but some do cut the Inner Starav Intrusion (408 Ma), including 589 lamprophyre (see above). Uncertainty as to when, after 408 Ma, the last dykes were 590 intruded causes some of the uncertainty in the overall duration of the Grampian post- 591 collision magmatic episode. This is expressed as ‘equal to or more than’ (≥) 22 592 million years, which takes initiation as ~430 and the ending at, or some time after, 593 ~408 Ma. 594 595 Morris et al. (2005) derived 40Ar – 39Ar ages of 415 ± 1.8 and 414 ± 2 Ma directly 596 from Etive dykes and concluded that they were emplaced at ~415 Ma, somehow 597 marking the end of magmatism and the Caledonian orogeny here. However, the full 598 age range of the dykes was not determined systematically from the field relationships 599 (e.g. Fig. 6) and the significantly younger age of the Inner Starav Intrusion, dated here 600 at 408 Ma (see above), was not recognised. Further, assignation of the magmatism to 601 the end of the Caledonian orogeny is distinctly problematic when neither the tectono- 602 magmatic setting nor the intended meaning of ‘orogeny’ is clarified. 603 604 An instructive feature of the Etive Dyke Swarm is that its northeastern limb extends 605 without deflection across the position of the previously influential northwest-trending 606 crustal discontinuity known as the Glencoe Lineament. This crust-penetrating fault 607 and shear zone (1) facilitated diverse magmas to rise rapidly to the surface at the 608 Glencoe caldera volcano, (2) repeatedly influenced caldera-collapse, and (3) 609 persistently formed a graben that directed a major river through the caldera (Moore 610 and Kokelaar 1997, 1998; Kokelaar and Moore 2006). To explain the lack of 611 deflection of the dykes, it is proposed that they propagated laterally northeastwards 612 from the Etive centre at shallow crustal levels. They were not influenced by the 613 Glencoe Lineament because they were emplaced within the solidified Clach Leathad 614 Pluton, which had obliterated the northwest-trending discontinuity in the upper crust. 615 The latter intrusion is more extensive at depth than at outcrop, extending outwards at

Neilson, Kokelaar, Crowley 18 616 least as far as beneath the Glencoe Fault-intrusions (Figs 3 and 5; British Geological 617 Survey 2005). 618 619 It is proposed that the Etive Dyke Swarm was directly involved in the growth of an 620 extensive and thick intermediate-to-silicic volcanic pile at and around the site 621 subsequently intruded by the Etive Pluton; we unimaginatively name this volcanic pile 622 – Etive Volcano. That there were vents and volcanic strata at the surface above the 623 dyke swarm seems intuitively obvious, but the considerable thickness of the volcanic 624 pile forming Etive Volcano has not previously been appreciated. In attempting to 625 reconcile the original 3-6 km of cover above the level of outcrop of the Etive Pluton 626 (Droop and Treloar 1981; Moazzen and Droop 2005; Droop and Moazzen 2007) with 627 the apparently shallow, 1-2 km-depth, dome-shaped and miarolitic roof-zone outcrop 628 of the neighbouring and earlier Clach Leathad Pluton, Kokelaar and Moore (2006) 629 suggested that the Etive Pluton may have been displaced upwards, by at least ~1.5-2 630 km, relative to the Clach Leathad Pluton. However, there is no evidence of a 631 continuous fault between the two intrusions, which are, in effect, stitched by unbroken 632 Etive dykes and by contact alteration of (earlier) truncated dykes (Fig. 6). 633 634 The evident difference between the original thicknesses of cover of the two plutons is 635 interpreted here as due to considerable thickening of the cover across the region after 636 emplacement of the Clach Leathad Pluton and before emplacement of the Cruachan 637 Intrusion. The greatest density of Etive dykes lies outside the Etive Pluton, implying 638 that many, if not the majority, predate it (Fig. 6; see also British Geological Survey 639 (2005) and the Geological Survey of Scotland (Oban) Sheet 45 (1907)). It is inferred 640 that the earlier Etive dykes, and probably a central vent complex (since obliterated), 641 built a substantial volcano south of the defunct Glencoe caldera volcano and Clach 642 Leathad Pluton. The thickness of the Etive Volcano pile would have had to have been 643 at least ~1.5-2 km. Such a thickness above the pre-existing Lorn Lava Pile, itself 644 ≥600 m thick and densely populated with Etive dykes, would account for the 645 greenschist-grade hydrothermal metamorphism of the Lorn lavas. The Etive pile, 646 which was entirely removed by erosion (the screen in the south of the Etive Pluton 647 may be a vestige of the Lorn pile), constituted ≥2000 km3 of intermediate-to-silicic 648 lavas and tuffs. Like Glencoe with its Clach Leathad Pluton, and Ben Nevis, Etive is 649 consistent with the relationship – volcano followed by pluton – supporting our

Neilson, Kokelaar, Crowley 19 650 hypothesis of similar patterns for many if not all ‘Newer Granite’ plutons, from 651 Donegal to Shetland. Etive demonstrates the point that, because of uplift and erosion, 652 intermediate-composition magmas have been grossly underestimated, with significant 653 implications for the overall petrogenesis. 654 655 Geochemistry 656 Southwest Grampian magmas, which formed for ≥22 million years after continental 657 collision, were dominated by intermediate to silicic compositions with great major- 658 element diversity (Fig. 2 a-c) and with considerable variations in the abundance of 659 incompatible trace elements (ITE: LILE, LREE, Nb, Zr and Y; e.g. Figs 2d and 7). 660 Most southwest Grampian magmas had the high Ba and Sr that characterises the 661 Argyll and Northern Highland plutonic suite, although low Ba-Sr types more 662 characteristic of the Cairngorm plutonic suite were also emplaced, both as plutons and 663 extrusions (Figs 2d and 7b). The plutons and volcanic successions show overlapped 664 compositions (e.g. Fig. 2a and d); the plutons are more restricted compositionally than 665 the extrusions and they present a greater proportion of silicic compositions at outcrop: 666 predominantly intermediate to silicic (granodiorite, monzogranite and granite) vs. 667 predominantly intermediate (andesite, quartz latite and dacite). Some of the relative 668 compositional restriction may reflect sampling fewer rocks from the plutons, although 669 it can be interpreted as due to a greater prevalence of hybridisation of coexisting 670 diverse magmas and remobilized cumulates during pluton growth. Key features, 671 which are mainly recognised from studies of the volcanoes, are (1) compositionally 672 diverse magma batches were emplaced together or separately in apparently haphazard 673 succession (Fig. 7b and c), and (2) there appears little correlation between pairs of ITE 674 (e.g. Figs 2d and 7b and c); binary ‘scattergrams’ indicative of highly variable ITE 675 ratios in different magma batches are characteristic. Strong tectonic control of 676 magma ascent and eruption, via crust-penetrating faults or shear zones, has been 677 established (e.g. Watson 1984; Moore and Kokelaar 1997, 1998) and, to explain the 678 key features, we suggest (1) both simultaneous and successive sampling of 679 compositionally diverse magmas that co-existed at depth, and (2) that many of the co- 680 existing magmas were not directly related genetically. Occurrence of numerous 681 distinct and not obviously related batches of magma was reported by Thirlwall (1979, 682 1981), especially for Lorn. 683 Fig. 7

Neilson, Kokelaar, Crowley 20 684 Magmas with primitive compositions were volumetrically scarce in the upper crust of 685 the southwest Grampian region. A few appinite-lamprophyre intrusions have Mg- 686 numbers ≥68 (100 Mg/(Mg+Fe2+)), but most rocks have ≤62. There are very few

687 magmatic bodies with SiO2 ≤50 wt.% (Fig. 2; Thirlwall 1979, 1981) and only a few,

688 amongst the Lorn lavas, with between 50 and 52 wt.% SiO2. Apart from the appinites, 689 no sample we have analysed has Ni >220 ppm or Cr >500 ppm, and thus it is inferred 690 that most of even the more mafic magmas that reached the upper crust were 691 significantly evolved relative to compositions of primary mantle melts. Thirlwall 692 (1981) considered that the moderately high Ni and Cr contents in the few basalts, and 693 in some basaltic andesites, indicated modified mantle melts with some cumulates. 694 695 The sampled appinite-lamprophyre intrusions include mafic and intermediate 696 compositions that indicate plagioclase-absent fractional crystallisation, consistent with 697 storage at depths ≥30 km in the presence of water (Neilson 2008). They generally 698 contain high Ni, Cr, V and MgO, coupled with elevated LILE and LREE abundances,

699 high Na2O + K2O, and K2O/Na2O generally ≥1. Figure 2d shows that most of the 700 sampled appinitic rocks maintain a fairly constant Sr/Ba ratio (~1.2), consistent with 701 their direct genetic relationship. 702 703 Roughly one third of the Lorn samples (33 out of 95) show compositional similarities 704 to the appinites-lamprophyres, in having several of the features: Ni >100 ppm, Cr 705 >200 ppm, V >100 ppm, Sr >1000 ppm, Ba >100 ppm, Rb >60 ppm, La >49 ppm, Ce

706 >79 ppm, Nd >35 ppm, Na2O + K2O >8 wt.%, and K2O/Na2O >0.95 (Fig. 7b; Neilson 707 2008); about one fifth are shoshonitic (17 samples, including 11 intrusions, have

708 SiO2≤57.0 wt.% and K2O≥Na2O-2.0; after Le Maitre 1989). Thus, superficially, it 709 appears possible that some of the Lorn magmas could be related to small-degree 710 mantle melts directly via fractional crystallisation. Similarly at Glencoe a few lavas 711 show appinite-lamprophyre affinities (Neilson 2008). However, in contrast to the 712 appinites and lamprophyres, most of the compositionally similar volcanic rocks 713 contain few phenocrysts, in many cases lacking hydrous phases (amphibole, 714 biotite/phlogopite) and including plagioclase. Importantly, in binary plots involving 715 incompatible trace elements, for example Figures 2d and 7b, the volcanic rocks 716 produce ‘scattergrams’ in which the majority of compositions cannot be related to the 717 appinitic series by crystal-liquid fractionation. The volcanic rocks contain the

Neilson, Kokelaar, Crowley 21 718 signature(s) of an ultimate origin in enriched lithospheric mantle, but most cannot be 719 related directly to the contemporary appinite-lamprophyre magmas. The geochemical 720 similarities are interpreted as reflecting a common ultimate source – the lithospheric 721 mantle – but with the extrusive and intrusive intermediate to silicic rocks mostly 722 derived by partial melting of the crust and not directly related to mantle melts by 723 fractional crystallisation. 724 725 Petrogenesis in the new time-frame 726 The timing of events in the Siluro-Devonian magmatic episode in the southwest 727 Grampian region, and the inferred concurrent tectonic processes, are summarised in 728 Figure 8. Episodes of voluminous eruption of diverse intermediate magmas were 729 common; the deeply eroded Grampian metamorphic basement was repeatedly covered 730 by extensive volcanic fields that mostly were eroded away quickly because of uplift, 731 for example, as recorded by contemporary conglomerates in the Midland Valley 732 (Bluck 1984). The Lorn Lava Pile (~426-425 Ma) represents a minimum of 300 km3 733 (probably >>1000 km3) of mainly andesite and dacite emplaced some 5 million years 734 before the Basal Andesite Sill-complex and its associated andesitic lavas at Glencoe 735 (≤420.5 Ma). Glencoe caldera volcano and its associated Fault-intrusions record 736 several further major eruptions of andesite and dacite, during ~420-418 Ma, and then, 737 during ~418-≤415 Ma, >2000 km3 of intermediate-to-silicic magmas were emplaced 738 to form the Etive Dyke Swarm with associated Etive Volcano. The several plutons, 739 Rannoch, Clach Leathad and Etive, also represent large volumes of intermediate to 740 silicic magmas. 741 Fig. 8 742 Mafic magmas that would form appinite-lamprophyre were certainly available for at 743 least 15 million years. Small volumes were emplaced in the upper crust together with 744 the intermediate to silicic magmas around and within the plutons, in the Lorn Lava 745 Pile and in the Etive Dyke Swarm (and hence also in Etive Volcano). Although both 746 appinite-lamprophyre and intermediate-silicic magmas utilised the same pathways to 747 the upper crust, as noted by Watson (1984), the compositions indicate that they were 748 not genetically related in the same way as in the hybrid appinite-syenite-granite 749 intrusions that record fractional crystallisation (e.g. Ach’uaine hybrid; Fowler 1988). 750 A relative volume problem arises (e.g. Tarney and Jones 1994; Fowler et al. 2001) in 751 attempting to attribute the large volumes of intermediate-silicic magmas to derivation

Neilson, Kokelaar, Crowley 22 752 by fractionation from appinite-lamprophyre parent magmas, albeit involving some 753 crustal assimilation. Fractional crystallisation alone to form huge volumes of andesite 754 and dacite from mantle-derived magma would require very substantial volumes of 755 mafic and ultramafic cumulates to form at or within the base of the crust, as in the 756 mafic sills of an active ‘deep crustal hot zone’ (see Annen et al. 2006). Despite 757 frequent tapping of diverse andesite-dacite magma batches along crust-penetrating 758 faults, little mafic magma was erupted or intruded and mafic cumulate xenoliths are 759 unknown. The substantial volume of intermediate to silicic magmas emplaced over 760 ≥22 million years is most easily reconciled with extensive partial melting of crust. 761 762 The considerable variability (characteristic scatter) of the ITE (e.g. Figs 2d and 7) is 763 interpreted as primarily reflecting partial melting of an inhomogeneous but mainly 764 incompatible-element-rich, mafic-to-intermediate lower crustal underplate and 765 fractional crystallisation of the mantle-derived appinitic-lamprophyric melts. Some 766 variability attributable to crustal assimilation is inevitable. The lower crustal melting, 767 at ≥950 °C (e.g. see Petford and Gallagher 2001; Annen et al. 2006) and in the 768 absence of plagioclase, would have needed a substantial source of heat and volatiles; it 769 also entailed uplift (see below). The persistent mantle-derived (appinite-lamprophyre) 770 magmas are interpreted as providing the required heat and volatiles. This new 771 petrogenetic model is illustrated in Figure 9. 772 Fig. 9 773 There are additional reasons for invoking mainly crustal melting together with some 774 fractionation of hydrous mantle melts. (1) Partial melting of mafic-to-intermediate 775 lower crust can simply account for the considerable variability of incompatible trace- 776 element contents, including the generation of low Ba-Sr magmas, via heterogeneity of 777 the crust and differing degrees of its melting. An origin primarily involving fractional 778 crystallisation would be difficult to reconcile with the compositional diversity and the 779 lack of any systematic compositional trend through time (Figs 2 and 7); it would have 780 to entail periodic tapping of numerous bodies of magma at widely differing stages of 781 fractional crystallisation, without accidental incorporation of cumulate. Presumably, 782 mafic restite is less easily incorporated in segregating partial melts, relative to the 783 cumulate fragments that are commonly included where fractional crystallisation has 784 occurred. (2) Appinitic fractionation series typically have strongly potassic

Neilson, Kokelaar, Crowley 23 785 intermediate daughter compositions (syenite-trachyte), rather than (trachy-) andesite. 786 In nature, extrusive counterparts of high-K calc-alkaline appinites and lamprophyres 787 are rare, although some shoshonitic lavas may fit this role (e.g. Rock 1991). In the 788 Grampian case, there are only a few lavas, in the Lorn pile, that are petrographically 789 and chemically closely similar to lamprophyres and are potassic (shoshonitic:

790 SiO2≤57.0 wt.%; K2O≥Na2O-2.0). (3) The high Ba-Sr link (e.g. Rock and Hunter 791 1987; Fowler and Henney 1996) and ‘volume problem’ are simply resolved if the 792 mafic-to-intermediate lower crust was formed from enriched sub-continental 793 lithospheric mantle during a previous episode of underplating. Mafic crust with 794 subduction-arc affinities was produced from the mantle at ~1.8 Ga (see above; 795 Marcantonio et al. 1988; Dickin and Bowes 1991; Muir et al. 1992, 1994) and this 796 seems a possible source in the later crustal melting. 797 798 The apparently minor occurrence of little-evolved lithospheric mantle melts (mafic 799 lamprophyres and appinite assemblages) in the upper crust does not preclude their 800 substantial presence at depth, where they would have provided a viable advective heat 801 and volatile source to partially melt the mafic-to-intermediate crustal underplate. In 802 early stages such magmas commonly reached the upper crust, but their ascent may 803 have become increasingly restricted by their greater density than the zones with 804 evolved magmas that formed above them. The model proposed here involves 805 repetitive intrusion of the lower crust by mantle melts throughout the magmatic 806 episode, conceivably reflecting the duration of availability of the asthenospheric heat 807 source. Repetitive intrusion of sheets of volatile-bearing mantle melt would be an 808 efficient mechanism in crustal melting (e.g. Petford and Gallagher 2001) and would 809 inevitably have led to some hybridisation of the respective magmas, contributing 810 further to the plethora of compositions in existence. Recent studies of oxygen isotopes 811 in zircons collected from dioritic rocks associated with ‘Newer Granites’ of the 812 Grampian Terrane (Loch Nagar pluton) also have indicated a petrogenesis primarily 813 involving crustal melting and the mixing of different batches of melts (Appleby et al. 814 2008). 815 816 The model proposed here involves both crustal growth, by addition from the 817 lithospheric mantle, and a considerable amount of crustal recycling (Fig. 9). The 818 magmatism followed the collision of Laurentia with Baltica and was initiated at the

Neilson, Kokelaar, Crowley 24 819 time of the collision of Laurentia with Avalonia. It is inferred that breakoff and 820 sinking of the subducted Iapetus lithospheric slab allowed buoyant ascent of ‘dry’, 821 relatively hot asthenosphere through the developing break to impact the hydrous, 822 enriched lithospheric mantle above (see also Atherton and Ghani 2002). The ascent of 823 asthenosphere must have been of sufficient scale and longevity to account for the ≥22 824 million-year duration of the magmatic episode, although it yielded no magma directly 825 (see Halliday et al. 1985), because it was dry. In cases of slab breakoff, the duration 826 of magmatism is related to the scale and longevity of the asthenospheric flow that is 827 caused by the sinking slab and which leads to heating and partial melting of the sub- 828 continental lithospheric mantle. In the European Alps, for instance, the duration of the 829 Tertiary slab-breakoff-related magmatism was ~17 million years. Shallower breakoff 830 leads to a larger anomaly and more melting (Davies and von Blankenburg 1995). 831 832 Discussion 833 As well as being impossible to reconcile with continuing active subduction, it is also 834 clear that the Grampian volcanoes were unlike typical subduction-related varieties 835 (e.g. Ewart 1976; Gill 1981). The Grampian andesites and dacites are relatively 836 phenocryst-poor, were erupted in unusually large volumes and with unusually low 837 viscosity (e.g. at Glencoe, Bidean nam Bian single cooling unit of ≥12 km3), and 838 predominantly formed laterally extensive lava sheets and sills; the ≥600 m-thick Lorn 839 pile has commonly been described as comprising plateau lavas (e.g. Pitcher 1982). 840 Big stratovolcanoes with broad volcaniclastic ring-plains are not recorded. Large 841 volumes of magma were tectonically facilitated to ascend rapidly to the surface, with 842 little opportunity for storage, fractionation, degassing or cooling at shallow levels 843 (Kokelaar and Moore 2006). 844 845 Our geochemical and field data together establish that the Grampian Siluro-Devonian 846 plutons and volcanic rocks are essential counterparts, the similar compositions of 847 which reflect essentially the same petrogenesis in this episode of post-collision 848 magmatism. Evidence that centred volcanoes were substantially obliterated by 849 genetically related plutons, and that substantial thicknesses of lavas and tuffs were 850 removed by erosion, much of it due to uplift during the magmatism (Kokelaar and 851 Moore 2006; Neilson 2008), leads us to advocate that similar volcanic activity 852 originally occurred widely across the entire Grampian belt of ‘Newer Granites’,

Neilson, Kokelaar, Crowley 25 853 including Donegal (e.g. Pitcher and Berger 1972) and Shetland (e.g. Flinn et al. 1968). 854 This proposal is consistent with the compositional overlap of the Grampian rocks and 855 Donegal plutons (Fig. 2a). 856 857 Most previous authors concerned with understanding the origin of the Caledonian 858 ‘Newer Granites’ have invoked either subduction or some relation to the uplift and 859 major strike-slip faulting that is known to have accompanied the magmatism. Those 860 advocating the latter (e.g. Pitcher 1982; Watson 1984) were influenced by those 861 ‘Newer Granites’ that occur in the Southern Uplands and England, clearly too close to 862 the Iapetus suture for them to be subduction-related. These more southerly plutons are 863 now recognised (Brown et al. 2008) as younger than those further north, and as 864 unrelated. This paper establishes that the Grampian magmatism was not related to on- 865 going subduction and that its onset at the time of continental collision strongly 866 suggests this as causative, as recognised by Atherton and Ghani (2002). Our model 867 explains the well known and distinctive association of appinite-lamprophyre magmas 868 with the high Ba-Sr granitoids and resolves the supposed ‘volume problem’ while 869 highlighting previously unrecognised huge volumes of intermediate magmas. Magma 870 chemistry, volumes and timing in relation to tectonism preclude explanations for the 871 magmatism in terms of lithosphere delamination or post-orogenic uplift-related 872 decompression melting. Slab breakoff is the best petrogenetic option available, 873 although it is impossible to prove unequivocally. 874 875 Atherton and Ghani (2002) were the first to advocate slab breakoff, to account for 876 ‘late granite syn-collisional magmatism’ in the Scottish Highlands and in Donegal. 877 They invoked progressive slab detachment that was initiated by collision between 878 Baltica and Laurentia and then produced a linear uplift belt with a singular, peaked 879 pulse of plutonic magmatism related to migration of the heat source (towards 880 Donegal). In the southwest Grampian study area no such peak is distinguished, but 881 instead a sustained episode involving repeated developments of volcanoes and plutons 882 over ≥22 million years within an area that has a length scale similar to the crustal 883 thickness. There is no sign of any protracted hiatus in the magmatic episode, or of any 884 protracted waxing or waning stage, although assimilation of silicic crust appears to 885 have increased towards the end (Neilson 2008), which is not surprising. Our model 886 also differs from that of Atherton and Ghani (2002) in recognising and accounting for

Neilson, Kokelaar, Crowley 26 887 the very large volumes of intermediate magmas that cannot be directly related to the 888 appinite-lamprophyre magmas. In their model, Atherton and Ghani (2002, figure 7) 889 first derive the appinitic magmas from the mantle to form a crustal underplate and then 890 partially melt this underplate to form the granites, whereas, for reasons detailed above, 891 we invoke persistent underplating and persistent melting of previously formed mafic 892 crust (Fig. 9). Atherton and Ghani’s model involves mainly crustal growth, whereas 893 the model presented here involves considerable crustal recycling. It is not clear 894 whether slab breakoff beneath the Grampian Terrane was a mechanical result of the 895 subduction-resistance of continental Avalonia, or due to lateral propagation of slab 896 detachment from the earlier collision involving Baltica. Slab segmentation may have 897 interrupted any lateral propagation of detachment and our data are inadequate to 898 resolve this uncertainty. 899 900 Uplift and erosion were rapid and considerable (evident from Ballachulish, Rannoch 901 Moor and Etive plutons; see above), with differential movements on major northwest- 902 and southwest-trending faults controlling the local preservation of thick volcanic 903 successions (Lorn and Glencoe) (Fig. 3). It is inferred that the uplift was stimulated 904 by early rebound consequent upon breakoff of the subducted dense lithospheric slab, 905 tumescence over the buoyantly rising asthenosphere, magmatic underplating, and 906 crustal heating. 907 908 It seems likely that a linear domain of asthenosphere rising through a slab break 909 would, as well as causing uplift, form a linear zone of melting of mantle and lower 910 crust that in turn would produce a linear belt of magmatism, thermal weakening, and 911 focussing of tectonic strain in the mid- and upper-crust (Fig. 9). The crust-penetrating 912 Great Glen Fault and similar parallel faults in the Grampian region (Fig. 3) would then 913 be consequent upon existence of the hot zone, even defining the dimensions of its 914 main locus of activity, with the crustal weakness being exploited on the switch to 915 plate-scale strike-slip. Both strike-slip and substantial normal displacements were 916 common on faults such as the Etive-Laggan, Ericht-Laidon, and Garabal 917 faults (Treagus 1991). The latter fault is inferred to have plumbed the early (~428-429 918 Ma) magmas of the Garabal Hill-Loch Fyne Complex (Rogers and Dunning 1991) and 919 the Etive-Laggan Fault, and its splays, clearly developed ≥500 m of downthrow to the 920 west shortly before facilitating major magma ascent during caldera volcanism at

Neilson, Kokelaar, Crowley 27 921 Glencoe (Kokelaar and Moore 2006). Logically one might expect the asthenospheric 922 plume to have broadened its influence through time, but our data are inadequate to test 923 for this. Our view, that uplift was consequent upon melt-forming processes and that 924 fault development and activity were focused by the magmatism, is the virtual opposite 925 of the former proposals (e.g. Pitcher 1982; Watson 1984) that the magmatism 926 passively resulted from uplift or was triggered from above by vertical and horizontal 927 fault-block movements. 928 929 JCN is grateful for NERC Postgraduate Studentship Award NER/S/A/2003/11285. We thank Adrian 930 Wood, Neil Boulton and Aaran Sumner for technical assistance at NIGL. The U-Pb dating was 931 supported by the NERC Isotope Facilities Steering Committee (Grant IP/794/1103 to BPK). We thank 932 Godfrey Fitton for providing XRF analytical facilities at Edinburgh and Kay Lancaster for cartographic 933 expertise at Liverpool. Matthew Thirlwall kindly contributed unpublished data for our plots of Etive 934 dyke compositions. Mike Fowler is thanked for a thorough review that improved the paper. Helen 935 Kokelaar compiled the references 936 937 References

938 Anderson, J. G. C. 1937. The Etive Granite Complex. Geological Society of London

939 Quarterly Journal, 93, 487-533.

940 Annen, C., Blundy, J. D. & Sparks, R. S. J. 2006. The genesis of intermediate and silicic

941 magmas in deep crustal hot zones. Journal of Petrology, 47, 505-539.

942 Appleby, S. K., Graham, C. M., Gillespie, M. R., Hinton, R. W., Oliver, G. H. & E.I.M.F.

943 2008. A cryptic record of magma mixing in diorites revealed by high-precision SIMS

944 oxygen isotope analysis of zircons. Earth and Planetary Science Letters, 269, 105-

945 117.

946 Atherton, M. P. & Ghani, A. A. 2002. Slab breakoff: a model for Caledonian, Late Granite

947 syn-collisional magmatism in the orthotectonic (metamorphic) zone of Scotland and

948 Donegal, Ireland. Lithos, 62, 65-85.

949 Bailey, E. B. 1960. The geology of Ben Nevis and and the surrounding country

950 (2nd edition). Memoir of the Geological Survey, Scotland, Sheet 53 (Scotland).

Neilson, Kokelaar, Crowley 28 951 Bailey, E. B. & Maufe, H. B. 1916. The geology of Ben Nevis and Glen Coe and the

952 surrounding country. Memoir of the Geological Survey, Scotland, Sheet 53 (Scotland).

953 Batchelor, R. A. & Bowden, P. 1985. Petrogenetic interpretation of granitoid rock series using

954 milticationic parameters. Chemical Geology, 48, 43-55.

955 Bluck, B. J. 1983. Role of the Midland Valley of Scotland in the Caledonian orogeny.

956 Transactions of the Royal Society of Edinburgh: Earth Sciences, 74, 119-136.

957 Bluck, B. J. 1984. Pre-Carboniferous history of the Midland Valley of Scotland. Transactions

958 of the Royal Society of Edinburgh: Earth Sciences, 75, 275-295.

959 Bluck, B. J. 2001. Caledonian and related events in Scotland. Transactions of the Royal

960 Society of Edinburgh: Earth Sciences, 91, 375-404.

961 Bowes, D. R. & Wright, A. E. 1967. The explosion-breccia pipes near Kentallen, Scotland,

962 and their geological setting. Transactions of the Royal Society of Edinburgh: Earth

963 Sciences, 67, 109-143.

964 British Geological Survey 2005. Glencoe. Bedrock. 1:25 000 Geology Series. (Keyworth,

965 Nottingham: British Geological Survey).

966 Brown, G. C. 1979. Geochemical and geophysical constraints on the origin and the evolution

967 of Caledonian granites. In: Harris, A. L., Holland, C. H. and Leake, B. E. (eds.) The

968 Caledonides of the British Isles - Reviewed. Geological Society, London, Special

969 Publications, 8, 645-652.

970 Brown, P. E., Ryan, P. D., Soper, N. J. & Woodcock, N. H. 2008. The Newer Granite

971 problem revisited: a transtensional origin for the Early Devonian Trans-Suture Suite.

972 Geological Magazine, 145, 235-256.

973 Davies, J. H. & von Blanckenburg, F. 1995. Slab breakoff - a model of lithosphere

974 detachment and its test in the magmatism and deformation of collisional orogens.

975 Earth and Planetary Science Letters, 129, 85-102.

Neilson, Kokelaar, Crowley 29 976 De la Roche, H., Leterrier, J., Grande claude, P. & Marchal, M. 1980. A classification of

977 volcanic and plutonic rocks using R1-R2 diagrams and major-element analyses - its

978 relationships with current nomenclature. Chemical Geology, 29, 183-210.

979 Dewey, J. F. 1971. A model for the Lower Palaeozoic evolution of the southern margin of the

980 early Caledonides of Scotland and Ireland. Scottish Journal of Geology, 7, 219-240.

981 Dewey, J. F. 2005. Orogeny can be very short. Proceedings of the National Academy of

982 Sciences, 102, 15286-15293.

983 Dewey, J.F. & Shackleton, R.M. 1984. A model for the evolution of the Grampian tract in the

984 early Caledonides and Appalachians. Nature, 312, 115-121.

985 Dewey, J. F. & Strachan, R. A. 2003. Changing Silurian-Devonian relative plate motion in the

986 Caledonides: sinistral transpression to sinistral transtension. Journal of the Geological

987 Society, London, 160, 219-229.

988 Dewey, J. F. & Strachan, R. A. 2005. The Caledonides of the British Isles. In: Selley, R. C.,

989 Cocks, L. R. M., and Plimer, I. (eds.) Encyclopedia of Geology. Academic Press, 56-

990 63.

991 Dickin, A. P. & Bowes, D. R. 1991. Isotopic evidence for the extent of early Proterozoic

992 basement in Scotland and northwest Ireland. Geological Magazine, 128, 385-388.

993 Droop, G. T. R. & Moazzen, M. 2007. Contact metamorphism and partial melting of

994 Dalradian pelites and semipelites in the southern sector of the Etive aureole. Scottish

995 Journal of Geology, 43, 155-179.

996 Droop, G. T. R. & Treloar, P. J. 1981. Pressures of metamorphism in the thermal aureole of

997 the Etive granite complex. Scottish Journal of Geology, 17, 85-102.

998 Ewart, A. 1976. Mineralogy and chemistry of modern orogenic lavas – some statistics and

999 implications. Earth and Planetary Science Letters, 31, 417-432.

1000 Flinn, D., Miller, J. A., Evans, A. L. & Pringle, I. R. 1968. On the age of the sediments and

Neilson, Kokelaar, Crowley 30 1001 contemporaneous volcanic rocks of western Shetland. Scottish Journal of Geology, 4,

1002 10-19.

1003 Fowler, M. B. 1988. Ach'uaine hybrid appinite pipes: evidence for mantle-derived shoshonitic

1004 parent magmas in Caledonian granite genesis. Geology, 16, 1026-1030.

1005 Fowler, M. B. & Henney, P. J. 1996. Mixed Caledonian appinite magmas: Implications for

1006 lamprophyre fractionation and high Ba-Sr granite genesis. Contributions to

1007 Mineralogy and Petrology, 126, 199-215.

1008 Fowler, M. B., Henney, P. J., Darbyshire, D. P. F. & Greenwood, P. B. 2001. Petrogenesis of

1009 high Ba-Sr granites: the Rogart pluton, . Journal of the Geological Society,

1010 London, 158, 521-534.

1011 Fraser, G. L., Pattison, D. R. M. & Heaman, L. M. 2004. Age of the Ballachulish and Glencoe

1012 Igneous Complexes (Scottish Highlands), and paragenesis of zircon, monazite and

1013 baddeleyite in the Ballachulish Aureole. Journal of the Geological Society, London,

1014 161, 447-462.

1015 Freeman, S. R., Butler, R. W. H., Cliff, R. A. & Rex, D. C. 1998. Dating mylonite evolution:

1016 an Rb-Sr and K-Ar study of the Moine mylonites, NW Scotland. Journal of the

1017 Geological Society, London, 155, 745-758.

1018 Furness, R. R. 1965. The petrography and provenance of the Coniston Grits east of the Lune

1019 Valley, Westmorland. Geological Magazine, 102, 252-260.

1020 Garnham, J. A. 1988. Ring-faulting and associated intrusions, Glencoe, Scotland. PhD thesis,

1021 University of London, 237pp.

1022 Ghani 1997. Petrology and geochemistry of the Donegal Granites, Ireland. PhD thesis,

1023 University of Liverpool, 204pp.

1024 Gibbons, W. & Gayer, R. A. 1985. British Caledonian Terranes. In: Gayer, R. A. (ed.), The

1025 Tectonic Evolution of the Caledonian-Appalachian orogen, Viewseg, Brunswick3-16.

Neilson, Kokelaar, Crowley 31 1026 Gill, J.B. 1981. Orogenic andesites and plate tectonics. Springer-Verlag. 390pp.

1027 Gradstein, F. M., Ogg, J. G. & Smith, A. G. 2004. A Geological Time Scale 2004. Cambridge

1028 University Press, Cambridge, 610pp.

1029 Halliday, A. N. 1984. Coupled Sm-Nd and U-Pb systematics in Late Caledonian granites and

1030 the basement under Northern Britain. Nature, 307, 229-233.

1031 Halliday, A. N., Stephens, W. E., Hunter, R. H., Menzies, M. A., Dickin, A. P. & Hamilton,

1032 P. J. 1985. Isotopic and chemical constraints on the building of the deep Scottish

1033 lithosphere. Scottish Journal of Geology, 21, 465-491.

1034 Harayama, S. 1992. Youngest exposed granitoid pluton on Earth - cooling and rapid uplift of

1035 the Pliocene-Quaternary Takidani granodiorite in the Japan Alps, Central Japan.

1036 Geology, 20, 657-660.

1037 Harmon, R. S., Halliday, A. N., Clayburn, J. A. P. & Stephens, W. E. 1984. Chemical and

1038 isotopic systematics of the Caledonian intrusions of Scotland and Northern England: a

1039 guide to magma source region and magma-crust interaction. Philosophical

1040 Transactions of the Royal Society of London, A310, 709-724.

1041 Heinz, W. & Loeschke, J. 1988. Volcanic clasts in Silurian conglomerates of the Midland

1042 Valley (Hagshaw Hills Inlier) Scotland, and their meaning for Caledonian plate

1043 tectonics. Geologische Rundschau, 77, 453-466.

1044 Highton, A. J. 1999. Introduction: Late Silurian and Devonian granitic intrusions of Scotland.

1045 In: Stephenson, D., Bevins, R. E., Millward, D., Highton, A. J., Parsons, I., Stone, P.

1046 and Wadsworth, W. J. (eds) Caledonian Igneous Rocks of Great Britain.

1047 Peterborough: Joint Nature Conservation Committee, 397-405.

1048 Hill, J. B. & Kynaston, H. 1900. On kentallenite and its relations to other igneous rocks in

1049 Argyllshire. Quarterly Journal of the Geological Society of London, 56, 531-558.

1050 Jacques, J. M. 1995. Caledonian magmatism and major tectonic structures in the SW

Neilson, Kokelaar, Crowley 32 1051 Highlands of Scotland: implications for ascent, siting and emplacement. PhD thesis,

1052 University of Durham, 479pp.

1053 Jacques, J. M. & Reavy, R. J. 1994. Caledonian plutonism and major lineaments in the SW

1054 Scottish Highlands. Journal of the Geological Society, London, 151, 955-969.

1055 Kaufmann, B. 2006. Calibrating the Devonian Time Scale: A synthesis of U-Pb ID-TIMS

1056 ages and conodont stratigraphy. Earth-Science Reviews, 76, 175-190.

1057 Kelling, G., Davies, P. & Holyroyd, J. 1987. Style, scale and significance of sand bodies in

1058 the Northern and Central Belts, southwest Southern Uplands. Journal of the

1059 Geological Society, London, 144, 787-805.

1060 Kemp, A. E. S. 1987. Evolution of Silurian depositional systems in the Southern Uplands,

1061 Scotland. In: Leggett, J. K. and Zuffa, G. G. (eds.) Marine Clastic Sedimentology.

1062 Graham & Trotman, 124-155.

1063 Kidston, R. & Lang, W. H. 1924. Notes on fossil plants from the Old Red Sandstone of

1064 Scotland. II. Nematophyton Forfarense, Kidston sp. III. On two species of Pachytheca

1065 (P. media and P. fasciculata) based on the characters of the algal filaments.

1066 Transactions of the Royal Society of Edinburgh, LIII, Part III (No. 28), 603-616.

1067 Kinny, P. D., Strachan, R. A., Kocks, H. & Friend, C. R. L. 2003. U-Pb geochronology of late

1068 Neoproterozoic augen granites in the Moine Supergroup, NW Scotland: dating of rift-

1069 related, felsic magmatism during supercontinent break-up? Journal of the Geological

1070 Society, London, 160, 925-934.

1071 Kneller, B. C. 1991. A foreland basin on the southern margin of Iapetus. Journal of the

1072 Geological Society, London, 148, 207-210.

1073 Kneller, B. C. & Bell, A. M. 1993. An Acadian mountain front in the English Lake District -

1074 the Westmorland monocline. Geological Magazine, 130, 203-213.

1075 Kneller, B. C., King, L. M. & Bell, A. M. 1993. Foreland basin development and tectonics on

Neilson, Kokelaar, Crowley 33 1076 the northwest margin of eastern Avalonia. Geological Magazine, 130, 691-697.

1077 Kocks, H., Strachan, R. A. & Evans, J. A. 2006. Heterogeneous reworking of Grampian

1078 metamorphic complexes during Scandian thrusting in the Scottish Caledonides:

1079 insights from the structural setting and U-Pb geochronology of the Strath Halladale

1080 Granite. Journal of the Geological Society, London, 163, 525-538.

1081 Kokelaar, B. P. 2007. Friction melting, catastrophic dilation and breccia formation along

1082 caldera superfaults. Journal of the Geological Society, London, 164, 751-754.

1083 Kokelaar, B. P. & Moore, I. 2006 Classical areas of British geology: Glencoe caldera

1084 volcano, Scotland. Keyworth, Nottingham, British Geological Survey.

1085 Krogh, T. E. 1982. Improved accuracy of U-Pb zircon dating by selection of more concordant

1086 fractions using a high-gradient magnetic separation technique. Geochimica et

1087 Cosmochimica Acta, 46, 631-635.

1088 Kynaston, H. & Hill, J. B. 1908. The geology of the country near Oban and Dalmally (Sheet

1089 45). Memoir of the Geological Survey, Scotland.

1090 Lee, G. W. & Bailey, E. B. 1925. The pre-Tertiary geology of Mull, Loch Aline and Oban.

1091 Memoir of the Geological Survey, Scotland. Parts of Sheets 35, 43, 44, 45 and 52

1092 (Scotland).

1093 Le Maitre, R.W. 1989. A classification of igneous rocks and glossary of terms. Blackwell,

1094 193pp.

1095 Leggett, J. K., McKerrow, W. S. & Eales, M. H. 1979. The Southern Uplands of Scotland; a

1096 lower Palaeozoic accretionary prism. Journal of the Geological Society, London, 136,

1097 755-770.

1098 Macdonald, R. & Fettes, D. J. 2007. The tectonomagmatic evolution of Scotland.

1099 Transactions of the Royal Society of Edinburgh: Earth Sciences, 97, 213-295.

1100 Macdonald, R., Rock, N. M. S., Rundle, C. C. & Russell, O. J. 1986. Relationships between

Neilson, Kokelaar, Crowley 34 1101 Late Caledonian lamprophyric, syenitic, and granitic magmas in a differentiated dyke,

1102 Southern Scotland. Mineralogical Magazine, 50, 547-557.

1103 Marcantonio, F., Dickin, A. P., McNutt, R. H. & Heaman, L. M. 1988. A 1,800-million-year-

1104 old Proterozoic gneiss terrane in Islay with implications for the crustal structure and

1105 evolution of Britain. Nature, 335, 62-64.

1106 Marshall, J. E. A. 1991. Palynology of the Stonehaven Group, Scotland: evidence for a Mid

1107 Silurian age and its geological implications. Geological Magazine, 128, 283-286.

1108 Moazzen, M. & Droop, G. T. R. 2005. Application of mineral thermometers and barometers

1109 to granitoid igneous rocks: the Etive Complex, W Scotland. Mineralogy and

1110 Petrology, 83, 27-53.

1111 Moore, I. & Kokelaar, P. 1997. Tectonic influences in piecemeal caldera collapse at Glencoe

1112 Volcano, Scotland. Journal of the Geological Society, London, 154, 765-768.

1113 Moore, I. & Kokelaar, P. 1998. Tectonically controlled piecemeal caldera collapse: A case

1114 study of Glencoe volcano, Scotland. Geological Society of America Bulletin, 110,

1115 1448-1466.

1116 Morris, G. A., Page, L. & Martinez, V. 2005. New dates (415 Ma) for the Etive Dyke Swarm

1117 and the end of the Caledonian Orogeny in the SW Grampian Highlands of Scotland.

1118 Journal of the Geological Society, London, 162, 741-744.

1119 Morton, D. J. 1979. Palaeogeographical evolution of the Lower Old Red Sandstone basin in

1120 the Western Midland Valley. Scottish Journal of Geology, 15, 97-116.

1121 Muir, R. J., Fitches, W. R. & Maltman, A. J. 1992. Rhinns Complex - a missing link in the

1122 Proterozoic basement of the North-Atlantic region. Geology, 20, 1043-1046.

1123 Muir, R. J., Fitches, W. R. & Maltman, A. J. 1994. The Rhinns Complex: Proterozoic

1124 basement on Islay and Colonsay, Inner Hebrides, Scotland, and on Inishtrahull, NW

1125 Ireland. Transactions of the Royal Society of Edinburgh: Earth Sciences, 85, 77-90.

Neilson, Kokelaar, Crowley 35 1126 Neilson, J. 2008. From slab breakoff to triggered eruptions: tectonic controls of Caledonian

1127 post-orogenic magmatism. PhD thesis, University of Liverpool, 518pp.

1128 Nockolds, S. R. 1941. Glen Fyne igneous complex. Quarterly Journal of the Geological

1129 Society of London, 106, 309-344.

1130 Oliver, G. J. H. 2001. Reconstruction of the Grampian episode in Scotland: its place in the

1131 Caledonian Orogeny. Tectonophysics, 332, 23-49.

1132 Oliver, G.J.H., Wilde, S.A. & Wan, Y. 2008. Geochronology and geodynamics of Scottish

1133 granitoids from the late Neoproterozoic break-up of Rodinia to Palaeozoic collision.

1134 Journal of the Geological Society, London, 165, 661-674.

1135 Pattison, D. R. M. & Harte, B. 1997. The geology and evolution of the Ballachulish igneous

1136 complex and aureole. Scottish Journal of Geology, 33, 1-29.

1137 Petford, N. & Gallagher, K. 2001. Partial melting of mafic (amphibolitic) lower crust by

1138 periodic influx of basaltic magma. Earth and Planetary Science Letters, 193, 483-499.

1139 Pidgeon, R.T. & Aftalion, M. 1978. Cogenetic and inherited zircon U-Pb systems in granites:

1140 Palaeozoic granites of Scotland and England. In: Bowes, D.R. and Leake, B.E. (eds)

1141 Crustal evolution in Northwestern Britain and adjacent regions. Geological Journal

1142 Special Issue, 10, 183-248.

1143 Pitcher, W.S. 1978. The anatomy of a batholith. Journal of the Geological Society, London,

1144 135, 157-182.

1145 Pitcher, W.S. 1982. Granite type and tectonic environment. In: Hsü, K.J. (ed.) Mountain

1146 Building Processes. Academic Press, London, 19–40.

1147 Pitcher, W.S. 1987. Granites and yet more granites forty years on. Geologische Rundschau,

1148 76, 51-79.

1149 Pitcher, W. S., Atherton, M. P., Cobbing, E. J. & Beckinsale, R. D. 1985. Magmatism at a

1150 Plate Edge: The Peruvian Andes. Blackie, Glasgow, 328pp.

Neilson, Kokelaar, Crowley 36 1151 Pitcher, W. S. & Berger, A. R. 1972. The geology of Donegal: a study of granite

1152 emplacement and unroofing. Wiley Interscience, New York.

1153 Read, H. H. 1961. Aspects of the Caledonian magmatism in Scotland. Proceedings of the

1154 Liverpool and Manchester Geological Society, 2, 653-683.

1155 Roberts, J. L. 1963. Source of the Glencoe ignimbrites. Nature, 5, 901.

1156 Rock, N. M. S. 1991. Lamprophyres. Blackie, 285pp.

1157 Rock, N. M. S. & Hunter, R. H. 1987. Late Caledonian dyke-swarms of Northern Britain -

1158 spatial and temporal intimacy between lamprophyric and granitic magmatism around

1159 the pluton, Inner Hebrides. Geologische Rundschau, 76, 805-826.

1160 Rogers, G. & Dunning, G. R. 1991. Geochronology of appinitic and related granitic

1161 magmatism in the W Highlands of Scotland - constraints on the timing of transcurrent

1162 fault movement. Journal of the Geological Society, London, 148, 17-27.

1163 Soper, N. J. 1986. The newer granite problem - a geotectonic view. Geological Magazine,

1164 123, 227-236.

1165 Soper, N.J., Strachan, R.A., Holdsworth, R.E., Gayer, R.A. & Greiling, R.O. 1992. Sinistral

1166 transpression and the Silurian closure of Iapetus. Journal of the Geological Society,

1167 London, 149, 871-880.

1168 Soper, N. J., Ryan, P. D. & Dewey, J. F. 1999. Age of the Grampian orogeny in Scotland and

1169 Ireland. Journal of the Geological Society, London, 156, 1231-1236.

1170 Soper, N. J. & Woodcock, N. H. 1990. Silurian collision and sediment dispersal patterns in

1171 Southern Britain. Geological Magazine, 127, 527-542.

1172 Stephens, W. E. 1988. Granitoid plutonism in the Caledonian orogen of Europe. In: Harris, A.

1173 L. and Fettes, D. J. (eds) The Caledonian-Appalachian Orogen. Geological Society,

1174 London, Special Publications, 38, 389-403.

1175 Stephens, W. E. & Halliday, A. N. 1984. Geochemical contrasts between late Caledonian

Neilson, Kokelaar, Crowley 37 1176 granitoid plutons of northern, central and southern Scotland. Transactions of the Royal

1177 Society of Edinburgh: Earth Sciences, 75, 259-273.

1178 Stephenson, D. & Gould, D. 1995. The Grampain Highlands. British Geological Survey,

1179 HMSO London, 261pp

1180 Stewart, M., Strachan, R. A. & Holdsworth, R. E. 1999. Structure and early kinematic history

1181 of the Great Glen Fault Zone, Scotland. Tectonics, 18, 326-342.

1182 Stewart, M., Strachan, R. A., Martin, M. W. & Holdsworth, R. E. 2001. Constraints on early

1183 sinistral displacements along the Great Glen Fault Zone, Scotland: structural setting,

1184 U-Pb geochronology and emplacement of the syn-tectonic Clunes tonalite. Journal of

1185 the Geological Society, London, 158, 821-830.

1186 Stone, P. 1995. Geology of the Rhinns of . Memoir of the British Geological Survey,

1187 HMSO London.

1188 Stone, P., Floyd, J. D., Barnes, R. P. & Lintern, B. C. 1987. A sequential back-arc and

1189 foreland basin thrust duplex model for the Southern Uplands of Scotland. Journal of

1190 the Geological Society, London, 144, 753-764.

1191 Tarney, J. & Jones, C. E. 1994. Trace-element geochemistry of orogenic igneous rocks and

1192 crustal growth models. Journal of the Geological Society, London, 151, 855-868.

1193 Thirlwall, M. F. 1979. The petrochemistry of the British Old Red Sandstone volcanic

1194 province. PhD thesis, University of Edinburgh, 314pp.

1195 Thirlwall, M. F. 1981. Implications for Caledonian plate tectonic models of chemical data

1196 from volcanic rocks of the British Old Red Sandstone. Journal of the Geological

1197 Society, London, 138, 123-138.

1198 Thirlwall, M. F. 1982. Systematic variation in chemistry and Nd-Sr isotopes across a

1199 Caledonian calc-alkaline volcanic arc - implications for source materials. Earth and

1200 Planetary Science Letters, 58, 27-50.

Neilson, Kokelaar, Crowley 38 1201 Thirlwall, M. F. 1986. Lead isotope evidence for the nature of the mantle beneath Caledonian

1202 Scotland. Earth and Planetary Science Letters, 80, 55-70.

1203 Thirlwall, M. F. 1988. Geochronology of Late Caledonian magmatism in Northern Britain.

1204 Journal of the Geological Society, London, 145, 951-967.

1205 Treagus, J. E. 1991. Fault displacements in the Dalradian of the Central Highlands. Scottish

1206 Journal of Geology, 27, 135-145.

1207 Trewin, N. H. & Thirlwall, M. F. 2002. Old Red Sandstone. In: Trewin, N. H. (ed.), The

1208 Geology of Scotland (4th edition). The Geological Society, London, 213-249.

1209 van Breemen, O. & Bluck, B. J. 1981. Episodic granite plutons in the Scottish Caledonides.

1210 Nature, London, 291, 113-117.

1211 Watson, J. 1984. The ending of the Caledonian orogeny in Scotland. Journal of the

1212 Geological Society, London, 141, 193-214.

1213 Weiss, S. & Troll, G. 1989. The Ballachulish igneous complex, Scotland: petrography,

1214 mineral chemistry and order of crystallization in the monzodiorite-quartz diorite suite

1215 and in the granite. Journal of Petrology, 30, 1069-1115.

1216 Wellman, C. H. 1994. Palynology of the 'Lower Old Red Sandstone' at Glen Coe, Scotland.

1217 Geological Magazine, 131, 563-566.

1218 Wellman, C. H. & Richardson, J. B. 1996. Sporomorph assemblages from the 'Lower Old Red

1219 Sandstone' of Lorne, Scotland. Special Papers in Palaeontology, 55, 41-101.

1220 Woodcock, N. H. & Strachan, R. A. 2002. The Caledonian Orogeny: a multiple plate

1221 collision. In: Woodcock, N. H. and Strachan, R. A. Geological history of Britain and

1222 Ireland. Blackwell, 187-206.

1223 Wright, A. E. & Bowes, D. R. 1979. Geochemistry of the Appinite Suite. In: Harris, A. L.,

1224 Holland, C. H. and Leake, B. E. (eds.) The Caledonides of the British Isles - Reviewed.

1225 Geological Society, London, Special Publications, 8, 699-704.

Neilson, Kokelaar, Crowley 39 1226 Zhou, J. -X. 1987. An occurrence of shoshonites near Kilmelford in the Scottish Caledonides

1227 and its tectonic implications. Journal of the Geological Society, London, 144, 699-

1228 706.

1229

Neilson, Kokelaar, Crowley 40 1230 Figure Captions

1231 Figure 1. The ‘Newer Granites’ of Scotland and Ireland (black). The intrusions in the 1232 Southern Uplands belong to the ‘Trans-Suture Suite’ attributed by Brown et al. (2008) 1233 to a transtensional origin; their ages range from 400-390 Ma, they have significant S- 1234 type characteristics and they are not petrogenetically related to the ‘Newer Granites’ 1235 further north that are the subject of this paper. SH is the Strath Halladale Granite; RC 1236 is the Rhinns Complex.

1237 Figure 2. (a) R1-R2 diagram for comparison of volcanic and plutonic rock 1238 compositions (De la Roche et al. 1980), showing tectono-magmatic discriminant fields 1239 applicable to granitoids (after Batchelor and Bowden 1985). Southwest Grampian 1240 pluton compositions (22 analyses) lie entirely within the Donegal ‘granites’ field 1241 (from Ghani 1997; Atherton and Ghani 2002), but barely overlap with the Coastal 1242 Batholith of Peru (Pitcher et al. 1985). Southwest Grampian volcanic compositions 1243 show considerable diversity (77 analyses from Glencoe, 95 from Lorn), with 1244 compositions on both sides of the ‘critical line of silica saturation’ (R1/R2=1). The 1245 Southwest Grampian pluton compositions lie almost entirely within the compositional 1246 field of the volcanics. Unlike the majority of Southwest Grampian volcanic and 1247 plutonic compositions, the Appinite Suite plots for the most part in the pre-plate 1248 collision field (13 analyses). (b) Southwest Grampian volcanic compositional field

1249 showing rock classification and contours of SiO2 wt.% (from De la Roche et al. 1980). 1250 (c) Detail of the Southwest Grampian volcanic compositions (triangles) and Etive 1251 Dyke Swarm compositions (diamonds: including data from Maufe and Bailey 1916; 1252 Garnham 1988; MF Thirlwall, unpublished). (d) Sr vs. Ba plot for Southwest 1253 Grampian volcanics, plutons, appinites and Etive Dyke Swarm. Such binary 1254 ‘scattergram’ plots of incompatible trace-element data are a distinctive feature of the 1255 magmatism, most simply reconciled with the occurrence of numerous batches of 1256 magma that had no direct genetic relationship. LowBaSr granitoid field from Tarney 1257 and Jones (1994).

1258 Figure 3. Southwest Grampian Siluro-Devonian post-collision magmatic rocks: 1259 plutons, volcanoes and dykes of the study region are shown with surrounding plutons 1260 and volcanoes. The Ben Nevis Dyke Swarm is not shown.

Neilson, Kokelaar, Crowley 41 1261 Figure 4. U-Pb concordia diagrams. (a) Lorne Lava Pile; (b) Rannoch Moor Pluton; 1262 (c) Glencoe Fault-intrusion; (d) Clach Leathad Pluton; (e) Cruachan Intrusion; (f) 1263 Inner Starav Intrusion.

1264 Figure 5. Regional map showing sample sites with the previous (unshaded boxes) and 1265 new U-Pb zircon ages (shaded boxes).

1266 Figure 6. Clach Leathad Pluton (pale green), dated at 418 Ma, is cut by the Meall 1267 Odhar and Cruachan Intrusions (reds) of the Etive Pluton, dated at 416-415 Ma (see 1268 text). Some Etive dykes (purple) are cut and metamorphosed by the Meall Odhar 1269 Intrusion, one dyke emanates from that intrusion and several dykes cut it and the 1270 Cruachan Intrusion. The outcrop of the older pluton is close to its original domed 1271 roof, formed at 1-2 km depth, whereas the Etive Pluton outcrop and wide 1272 metamorphic aureole (not shown) indicate significantly thicker original cover, perhaps 1273 3-6 km (Droop and Treloar 1981; Moazzen and Droop 2005; Droop and Moazzen 1274 2007). Vertical relative displacement between the plutons is precluded by the dykes 1275 that (1) are truncated and altered by the Meall Odhar Intrusion, and (2) emanate from 1276 or cut straight across the contacts. The figure is based on original Geological Survey 1277 mapping (reported by Bailey 1960) compiled in a new map (British Geological Survey 1278 2005). FI is Fault-intrusion (of Glencoe); DAQ and DAS are Dalradian quartzites and 1279 schists, respectively. Map coordinates relate to Ordnance Survey 100 km square NN.

1280 Figure 7. (a) Harker plots of representative incompatible trace-element compositions 1281 in Southwest Grampian volcanic, plutonic, appinite-lamprophyre and dyke rocks 1282 (Neilson 2008). (b) Compositional variation in the Lorn Lava Pile and associated 1283 intrusions, demonstrating that despite a wide range of states of compositional

1284 ‘evolution’, here in MgO vs. SiO2, there is no evolutionary trend through time, 1285 expressed as MgO vs. stratigraphic height in the ≥600 m succession (intrusions are 1286 plotted at height equals zero). Widely different magma batches were emplaced 1287 successively and haphazardly during ≤1 million years, with widely varying ratios of 1288 incompatible trace elements (here Sr and Ba). Such ‘scattergram’ plots involving 1289 incompatible trace elements are a general characteristic of the magmatism (e.g. Fig. 1290 2d). (c) Contrasting compositional batches of magma emplaced together in a single 1291 caldera-wide lake of ≥12 km3 of magma erupted via crust-penetrating faults. The 1292 samples derive from four vertical transects spaced across Glencoe caldera volcano; the

Neilson, Kokelaar, Crowley 42 1293 rocks constitute a single cooling unit, locally ≥350 m thick, and no internal contacts 1294 are discernible (Kokelaar and Moore 2006; Neilson 2008).

1295 Figure 8. Time-lines for Siluro-Devonian tectonic and magmatic activity relevant to 1296 the southwest Grampian region. The Silurian-Devonian boundary is at 418 Ma 1297 (Kaufmann 2006). Age ranges of Ballachulish and Rannoch Moor plutons are less 1298 certain than others. Etive Volcano and Etive Dyke Swarm were mostly formed during 1299 the first ~4 million years; the extensive ranges of these account for the few dykes, 1300 including lamprophyre, that cut the Inner Starav Intrusion dated at 408 Ma (see text). 1301 1Onset age derived from recalculation of results of Rogers and Dunning (1991). 1302 2After Fraser et al. (2004). 3After Dewey and Strachan (2005).

1303 Figure 9. Schematic diagram illustrating the proposed petrogenetic model, involving 1304 slab breakoff and consequent rise of relatively hot and dry asthenosphere, partial 1305 melting of LILE- and LREE-enriched lithospheric mantle, new crustal addition of 1306 magmas that form appinite-lamprophyre, and crustal recycling by partial melting of 1307 mafic-to-intermediate lower crust to form plutons and volcanoes. Ascent of diverse 1308 magmas to the upper crust was facilitated by major faults that formed in this region as 1309 a consequence of the thermal weakening of the crust here and the onset of plate-scale 1310 sinistral strike-slip. Considerable uplift resulted from the slab detachment, from the 1311 buoyant rise of asthenosphere, and from the heating of the lithosphere. The model 1312 shows a putative early to middle stage of the magmatic episode and it is inferred that 1313 the asthenospheric plume broadens through time.

1314

Neilson, Kokelaar, Crowley 43

0.0696 data-point error ellipses are 2σ 0.0680 data-point error ellipses are 2σ Mean 206Pb/238U age = 425.0±0.6 95% conf. Mean 206Pb/238U age = 422.7±0.3 95% conf. Wtd by data-pt errs only, 0 of 2 rej. Wtd by data-pt errs only, 0 of 4 rej. MSWD = 2.5, probability = 0.11 432 MSWD = 0.67, probability = 0.57 0.0692 0.0679

423 0.0688 U 0.0678 U 238

238 428 Pb/ Pb/ 206 0.0684 0.0677 206 422

Concordia Age = 425.0 ±0.7 Ma Concordia Age = 422.5 ±0.5 Ma 0.0680 424 0.0676 (2σ, decay-const. errs included) (2σ, decay-const. errs included) MSWD (of concordance) = 0.025, MSWD (of concordance) = 1.3, (a) Probability (of concordance) = 0.87 (b) Probability (of concordance) = 0.26 0.0676 0.0675 0.514 0.518 0.522 0.526 0.530 0.534 0.5135 0.5145 0.5155 0.5165 0.5175 207 235 207 235 Pb/ U Pb/ U data-point error ellipses are 2σ data-point error ellipses are 2σ 0.0682 Intercepts at Mean 206Pb/238U age = 417.8 ±0.9 95% conf. 0.0673 Wtd by data-pt errs only, 0 of 5 rej. 0.0678 206±260 & 419.6±5.4 Ma MSWD = 3.5, probability = 0.01 MSWD = 1.01 422

0.0674 0.0671

U 418

0.0670 U 418 238 238 Pb/ 0.0666 Pb/ 0.0669 206 206 414 0.0662

0.0667 Concordia Age = 417.9 ±0.9 Ma 416 (95% confidence, decay-const. errs included) 0.0658 410 MSWD (of concordance) = 0.30, Probability (of concordance) = 0.58 (c) (d) 0.0654 0.0665 0.44 0.46 0.48 0.50 0.52 0.54 0.56 0.506 0.507 0.508 0.509 0.510 0.511 0.512 207 235 Pb/ U 207Pb/235U data-point error ellipses are 2σ 0.067 data-point error ellipses are 2σ 418 Mean 206Pb/238U age = 414.7±0.7 95% conf. Mean 206Pb/238U age = 408.0 ±0.3 95% conf. Wtd by data-pt errs only, 0 of 4 rej. 416 Wtd by data-pt errs only, 0 of 3 rej. 0.0665 MSWD = 3.9, probability = 0.01 MSWD = 0.11, probability = 0.89

0.066 410 U U 0.0655 238

238 408 0.065 Pb/ Pb/ 206 206 402 0.0645 0.064 Concordia Age = 414.9 ±0.7 Ma Concordia Age = 408.0 ±0.5 Ma (95% confidence, decay-const. errs included) (2σ, decay-const. errs included) MSWD (of concordance) = 0.13, MSWD (of concordance) = 0.053, Probability (of concordance) = 0.82 (e) Probability (of concordance) = 0.72 (f) 0.0635 0.063 0.484 0.488 0.492 0.496 0.500 0.504 0.508 0.475 0.485 0.495 0.505 207 235 207 235 Pb/ U Pb/ U

Figure 4. U-Pb concordia diagrams. A: Lorne Lava Pile; B: Rannoch Moor Pluton; C: Glencoe Fault Intrusion; D: Clach Leathad Pluton; E: Cruachan Intrusion; F: Inner Starav Intrusion.

1 Analytical Techniques 2 Zircon fractions were analysed by Isotope Dilution Thermal Ionization Mass 3 Spectrometry (ID-TIMS) at the NERC Isotope Geosciences Laboratory (NIGL), 4 Nottingham, UK and at the Jack Scatterly Geochronology Laboratory, University of 5 Toronto. Analytical procedures were largely as described in Noble et al. (1993). 6 Zircon crystals were separated from <355 µm bulk-rock powder using standard 7 vibrating-table, specific gravity and magnetic techniques. Crystal fractions for 8 analysis were picked under a binocular microscope. Initially a batch of zircons was 9 air abraded (Krogh 1982), but subsequently a modified chemical abrasion technique 10 was used. Bulk zircon fractions were annealed at between 800 and 900 oC in quartz 11 glass beakers for between 48 and 60 hours. The zircon crystals were ultrasonically

12 washed in 4N HNO3, rinsed in ultra-pure water, then further washed in warm 4N

13 HNO3 prior to rinsing with distilled water to remove surface contamination. The 14 annealed, cleaned bulk zircon fractions were then chemically abraded in 200 µl 29N o 15 HF and 20 µl 8N HNO3 between 120 and 180 C for 12 hours to minimise or 16 eliminate Pb-loss (technique modified after Mattinson 2005). Chemically abraded 17 zircons were washed several times in ultra-pure water, cleaned in warm 3N HCl for

18 several hours on a hot-plate, rinsed again in ultra-pure water and 8N HNO3 and split 19 into single grain fractions ready for dissolution. Mixed 205Pb – 235U or 205Pb – 233U – 20 235U tracers were used to spike all fractions. The first batch of zircons, which was air 21 abraded, underwent chemical separation procedures prior to mass spectrometry. 22 Subsequent chemically abraded batches were not subjected to ion-exchange 23 procedures, but were converted to chloride and loaded onto degassed rhenium 24 filaments in silica gel, following a procedure modified after Mundil et al. (2004). 25 Isotope data at the Jack Scatterly Geochronology Laboratory were collected using a 26 VG-354 equipped with a Daly ion-counter, whereas at NIGL a Thermo Electron 27 Triton equipped with a new generation of MassCom Secondary Electron Multiplier 28 was utilised (Noble et al. 2006). A minimum of 100 ratios were collected for Pb and 29 60 for U. Pb ratios were scrutinised for any evidence of organic interferences, which 30 were determined to be negligible. Errors were calculated using numerical error 31 propagation (Ludwig 1980). Isotope ratios were plotted using Isoplot version 3 32 (Ludwig 1993, 2003); error ellipses reflect 2σ uncertainty. Total procedural blanks 33 for three separate batches of chemistry between October 2004 and April 2006 were 34 2.0 to 0.2 pg for Pb and 0.3 to 0.1 pg for U. Samples were blank corrected using the 35 measured blank 204Pb:206Pb:207Pb:208Pb ratio of 1:18.70:15:15:36.82. Correction for 36 residual common lead above analytical blank was carried out using the Stacey- 37 Kramers common lead evolutionary model (Stacey and Kramers 1975). 38 39 References 40 Krogh, T. E. 1982. Improved accuracy of U-Pb zircon dating by selection of more

41 concordant fractions using a high-gradient magnetic separation technique.

42 Geochimica et Cosmochimica Acta, 46, 631-635.

43 Ludwig, K. R. 1980. Calculation of uncertainties of U-Pb isotope data. Earth and

44 Planetary Science Letters, 46, 212-220.

45 Ludwig, K. R. 1993. Isoplot: a plotting and regression program for radiogenic isotope

46 data, Version 2.70, US Geological Survey Open File Report, 91-445.

47 Ludwig, K. R. 2003. User's Manual for Isoplot 3.00, Berkeley Geochronology Center,

48 Berkeley, California.

49 Mattinson, J. M. 2005. Zircon U-Pb chemical abrasion ("CA-TIMS") method:

50 Combined annealing and multi-step partial dissolution analysis for improved

51 precision and accuracy of zircon ages. Chemical Geology, 220, 47-66.

52 Mundil, R., Ludwig, K. R., Metcalfe, I. & Renne, P. R. 2004. Age and timing of the

53 Permian mass extinctions: U/Pb dating of closed-system zircons. Science, 305,

54 1760-1763.

55 Noble, S. R., Schweiters, J., Condon, D. J., Crowley, Q. G., Quaas, N. & Parrish, R.

56 R. 2006. TIMS characterization of new generation of secondary electron

57 multiplier. EOS Transactions of the American Geophysical Union, 87 (52),

58 Fall Meeting Supplement, Abstract.

59 Stacey, J. S. & Kramers, J. D. 1975. Approximation of terrestrial lead isotope

60 evolution by a 2-stage model. Earth and Planetary Science Letters, 26, 207-

61 221. 62 Supp Captions 63 64 Supplementary Table 1. U-Pb Isotope Dilution Thermal Ionisation Mass 65 Spectrometry (ID-TIMS) data 66 67 Supplementary Figure 1. Cathode luminescence images of zircons from Lorn 68 dacite sample JNL86. (a) Representative of the analysed acicular zircons with 69 simple magmatic zoning, c-axis magmatic inclusions, and no inherited cores. 70 (b) Representative of the distinctly more equant zircons with magmatic 71 overgrowth on inherited cores, which were excluded from analysis.

72 Supplementary Figure 2. Textures due to mingling of appinitic and 73 monzogranitic magmas in the Clashgour intrusion (NN 228 434). Lobate to 74 crenulated contacts and streaky to irregular-shaped inclusions distinguish 75 coexistence as fluids, probably crystal-bearing, while rather ‘blurred’ contacts, 76 labelled H, indicate local hybridisation.

77

Table 1. Sample Fraction Weight U Cm-Pb Ratios (µg) (ppm) (ppm) ‡ 206Pb/204Pb † 207Pb/206Pb* 2σ% 206Pb/238U* 2σ% JNL86 - Lorn Z1 1.0 328 4.5 210.87 0.05550 0.24 0.06907 0.18 Z2 4.1 243 1.6 895.18 0.05552 0.14 0.06824 0.19 Z5 7.6 90 3.3 482.95 0.05503 0.14 0.06895 0.17 Z6 4.5 163 4.9 408.31 0.05541 0.14 0.06805 0.19 JNL88 - Z5 1.1 1095 0.8 1463.03 0.05519 0.06 0.06769 0.17 Rannoch Moor Z6 3.2 635 0.7 2468.14 0.05520 0.06 0.06781 0.16 Z8 0.6 1820 1.0 946.35 0.05517 0.06 0.06776 0.16 Z9 0.3 885 3.1 647.62 0.05520 0.06 0.06778 0.16 JNGC76 - Z1 1.7 125 2.5 91.41 0.05503 7.81 0.06618 0.53 Glencoe Fault Intrusion Z2 1.6 149 8.7 1639.34 0.05517 0.86 0.06636 0.14 Z3 2.1 962 12.5 900.90 0.05502 1.08 0.06660 0.16 Z4 9.4 193 6.9 1732.79 0.05515 1.06 0.06608 0.13 Z5 1.1 290 10.5 1661.72 0.05514 0.68 0.06686 0.12 Z6 3.9 307 3.3 6587.27 0.05516 1.52 0.06672 0.16 Z7 1.1 437 0.3 6582.16 0.05514 0.09 0.06648 0.21 Z8 1.1 373 8.1 232.64 0.05529 0.22 0.06755 0.17 Z9 1.3 509 7.2 401.67 0.05506 0.15 0.06681 0.19 Z10 1.1 415 7.8 259.28 0.05513 0.21 0.06596 0.20 JNE5 - Clach Leathad Z27 2.8 524 0.9 6731.17 0.05519 0.14 0.06701 0.20 Z28 0.6 1025 1.3 2061.83 0.05521 0.09 0.06690 0.17 Z29 1.7 309 0.7 3097.45 0.05518 0.13 0.06683 0.20 Z30 3.3 313 1.1 4102.95 0.05513 0.07 0.06710 0.16 Z31 3.0 245 1.9 1630.51 0.05519 0.08 0.06686 0.17 JNE8 - Cruchan Z33 8.9 761 66.4 361.65 0.05497 0.57 0.06632 0.18 Z34 4.0 603 1.7 6725.00 0.05512 0.22 0.06447 0.16 Z35 8.0 550 4.4 4045.31 0.05508 0.12 0.06458 0.22 Z36 7.0 291 2.8 3018.55 0.05506 0.07 0.06648 0.06 Z37 6.3 468 1.8 7028.43 0.05510 0.06 0.06644 0.16 Z38 10.6 3 21.5 4097.55 0.05508 0.07 0.06634 0.16 Z4 1.4 254 0.4 3439.20 0.05507 0.12 0.06370 0.16 Z5 2.5 489 2.2 2378.29 0.05540 0.16 0.06600 0.17 JNGC87 - Starav Z10 7.1 350 3.3 14055.72 0.05500 0.75 0.06534 0.14 Z41 8.4 452 3.8 12397.28 0.05496 0.50 0.06534 0.16 Z42 4.5 175 0.9 3819.94 0.05509 0.10 0.06531 0.19

All errors are 2σ (per cent for ratios, absolute for ages) ‡ Total common Pb in analysis, corrected for spike and fractionation (0.09%/amu) † Measured ratio, corrected for spike and Pb fractionation * Corrected for blank Pb, U and common Pb (Stacey and Kramers 1975) Ratios Ages (Ma)

207Pb/235U* 2σ% Rho 207Pb/206Pb 2σ abs 206Pb/238U 2σ abs 207Pb/235U 2σ abs 0.52849 0.31 0.66 432.3 0.4 430.5 0.8 430.8 1.3 0.52011 0.24 0.81 433.0 3.1 425.4 0.8 426.8 1.0 0.52759 0.22 0.78 413.6 3.1 429.6 0.7 427.1 0.9 0.51921 0.24 0.81 428.8 3.2 424.5 0.8 425.1 1.0 0.51516 0.18 0.94 420.0 1.3 422.2 0.7 421.9 0.8 0.51609 0.17 0.93 420.0 1.4 422.9 0.7 422.5 0.7 0.51540 0.17 0.93 419.1 1.4 422.6 0.7 422.1 0.7 0.51589 0.18 0.91 420.3 1.2 422.8 0.7 422.4 0.7 0.50219 7.82 0.31 413.6 1.8 413.1 2.1 413.2 26.9 0.50481 0.92 0.48 419.2 19.2 414.2 0.6 415.0 3.1 0.50523 1.14 0.42 413.1 24.0 415.6 0.7 415.2 3.9 0.50224 1.14 0.60 418.2 23.7 412.5 0.5 413.4 3.9 0.50833 0.73 0.45 417.9 15.2 417.2 0.5 417.3 2.5 0.50741 1.62 0.69 418.6 33.5 416.4 0.7 416.7 5.5 0.50547 0.23 0.93 418.0 1.9 414.9 0.9 415.4 1.0 0.51495 0.28 0.64 423.8 4.8 421.4 0.7 421.8 1.2 0.50721 0.24 0.80 414.7 3.3 416.9 0.8 416.6 1.0 0.50140 0.30 0.70 417.7 4.7 411.8 0.8 412.7 1.2 0.50991 0.24 0.80 419.8 3.2 418.1 0.8 418.4 1.0 0.50924 0.19 0.89 420.6 1.9 417.5 0.7 417.9 0.8 0.50846 0.24 0.84 419.5 2.9 417.1 0.8 417.4 1.0 0.51011 0.18 0.93 417.7 1.5 418.7 0.7 418.5 0.8 0.50873 0.18 0.91 419.8 1.7 417.2 0.7 417.6 0.8 0.50268 0.62 0.39 411.2 12.7 413.9 0.7 413.5 2.6 0.48991 0.27 0.61 417.0 4.8 402.7 0.6 404.9 1.1 0.49048 0.22 0.88 415.7 2.7 403.4 0.9 405.2 0.9 0.50469 0.18 0.92 414.7 1.5 414.9 0.2 414.9 0.7 0.50481 0.17 0.94 416.4 1.3 414.7 0.7 414.9 0.7 0.50383 0.18 0.93 415.7 1.5 414.0 0.7 414.3 0.7 0.48181 0.18 0.90 414.8 2.8 398.5 0.6 399.4 0.7 0.50282 0.35 0.94 428.4 3.6 412.4 0.7 413.3 1.5 0.49554 0.81 0.50 412.3 16.8 408.0 0.5 408.7 2.7 0.49515 0.55 0.48 410.7 11.0 408.0 0.6 408.4 1.9 0.49427 0.34 0.60 415.7 2.3 407.8 0.8 407.8 1.4